Difficult to Treat Infections in the Burn Patient

Osteomyelitis

Osteomyelitis is very uncommon among burn patients. An incidence of 2% is reported in retrospective case series in patients with severe burns with the majority of data coming from pediatric patients. Data are lacking on deaths in burn patients with osteomyelitis. In addition, bone and joint changes secondary to burns are often not defined [1,2].

Predisposing factors for osteomyelitis include the extent of the burn surface area and burn depth, together with exposure of musculoskeletal structures beneath the involved skin, although the pathogenesis is different in burn patients than in healthy hosts. Deep burns with exposure of underlying musculoskeletal tissue and electrical burns can lead to thrombosis in the peri-osteal nutrient vessels in bone producing tissue necrosis that allows for contiguous infection.

The management of osteomyelitis includes a combined medical and surgical approach. Dead tissue must be removed because poorly vascularized tissue is unlikely to heal. Clinical signs of osteomyelitis may be non-specific and difficult to recognize. The presence of exposed bone, persistent sinus tract and tissue necrosis overlying bone should raise suspicion for osteomyelitis. Laboratory abnormalities including leukocytosis and elevated inflammatory markers are non-specific indicators particularly in a critically ill burn patient.

Imaging is useful, but sensitivity and specificity vary depending on the modality used (Table 1). Magnetic resonance imaging (MRI) is considered the imaging of choice, detecting early osteomyelitis as well as necrotic bone, sinus tracts, and abscesses [3,4]. Early recognition with surgical debridement and bone biopsy for culture and directed antimicrobial treatment are imperative to prevent related functional sequelae and deformities [1,5]. Skeletal fixation should be considered in severely burned patients if indicated because it appears to present a low risk for infectious complications [6].

A retrospective review of 600 children (ages one month to 18 years) hospitalized in a pediatric burn unit over a 10-year period found 12 children received a diagnosis of osteomyelitis [2]. The median age and total body surface area (TBSA) was 42.5 months and 33.5%, respectively, with mechanism of direct fire in 11 patients and electrical burn in one. Eleven patients had bone exposure on admission and seven had compartment syndrome of the affected site. Osteomyelitis was diagnosed at a median of 30 days post-burn. Upper extremity was the most common site of involvement as has been reported in another series [1].

Fungal osteomyelitis was the most common etiology (in nine of the 12 cases) and was attributed to prolonged use of broad-spectrum antibiotic agents, large TBSA involvement, and presence of venous and arterial lines. Pseudomonas aeruginosa (two patients) and Enterococcus faecalis (one patient) were isolated from the other patients. In addition, two patients had a mixed infection with Candida parapsilosis and Acinetobacter baumannii from one and C. albicans and P. aeruginosa from another. Complications were seen in 11 patients; six had motor sequelae and four underwent amputation of the affected site. Reportedly, one patient died, with death attributed to osteomyelitis-related sepsis [2]. While details are lacking, the patient was a 48-month-old child (17% TBSA involvement) with C. albicans osteomyelitis of the tibia.

Recurrent and Polymicrobial Blood Stream Infections

Severe thermal injury, with its hyperdynamic physiology and immunomodulatory effects, prolonged hospitalization, and multiple invasive devices and procedures, increases risk for BSI and complexity of management. This is manifested frequently as sepsis, so we refer the reader to the accompanying article on this topic. In this section, we will review the epidemiology and challenges associated with BSI in the burn unit, and then concentrate on recurrent and polymicrobial BSI.

The prevalence and etiology of BSI has changed with the evolving management of burns. Burn centers that adopted early debridement and improved infection control practices noted bacteremia-related mortality secondary to burn wound infection dropping from 40% to 18% [7]. While wound infections still contribute to bacteremia, there is growing evidence for other nosocomial sources, especially catheter-related BSI (CRBSI), which have risen from 15% in the 1980s to 27%35% in recent studies [8–10].

Compared with other intensive care units (ICUs), with predominantly gram-positive BSIs, 63%–92% of BSIs in burns are because of gram-negative organisms. Staphylococcus aureus is the predominant gram-positive in burn, nosocomial, and surgical ICU data [8,9,11–14]. While P. aeruginosa is common across nosocomial, surgical, and burn ICU-associated gram-negative BSIs, A. baumannii and Klebsiella pneumoniae are among the most frequent gram-negative organisms in burns (compared with Escherichia coli in other nosocomial BSIs) [8,9,14,15]. Health-care–associated infections in burn patients evolve from gram-positive to increasingly resistant gram-negative organisms by four weeks of hospitalization [16–18].

Polymicrobial BSI presents an additional challenge in the burn patient. Studies among non-burn hospitalized patients have found a prevalence of 6.1%–19.5% [15,19,20]. A retrospective study of 58 patients with a mean TBSA of 40% from the Formosa Fun Coast dust explosion (where a majority of patients submerged themselves after injury) found that 18.2% of BSIs were polymicrobial at an average of 18 days after admission [12]. An eight-year retrospective review of nosocomial BSIs in a burn unit found that 12.1% of their CRBSIs were polymicrobial [21]. Depending on the setting of the burn, polymicrobial BSIs are predominantly gram-negative or Candida spp [8,9,11,19].

Recurrent BSI is defined as BSI that occurs with the same organism after initial culture clearance [9,10]. In prospective reviews of non-burn hospitalized patients, 7.1% and 9.4% had recurrent gram-negative and S. aureus BSI, respectively [22]. While A. baumannii is the predominant pathogen in monomicrobial initial episodes of BSI, in recurrences, P. aeruginosa (36%–57%) and K. pneumoniae (46%) are the most common organisms [8–11].

There are limited data on recurrent BSI in the burn population; a retrospective study of burn patients over 10 years, with a median burn TBSA of 35% and cumulative incidence of BSI of 4.4 episodes/100 patients, found that 6.8% of patients had recurrent BSI, which was predominantly P. aeruginosa [9]. A military study of combat-related burn patients with a median burn TBSA of 41% and high incidence of perineal burns found a recurrence rate of 37% for all BSIs with a median of 20 days to recurrence. Risk factors identified for recurrence were TBSA >20%, perineal burns, urinary tract infection as the etiology of BSI, and longer durations of hospitalization and mechanical ventilation. In this study of combat-related burn patients, 50% of the recurrences were polymicrobial [10]. This may be difficult to extrapolate to the non-combat population because of the mechanisms of burn injury, potential for delayed presentation because of the combat environment, and other associated injuries.

In a review of patients from a burn ICU (BICU) from 1980 to1986, mortality rate was 28% and 4.3% in those with and without BSI, respectively. Increased death was associated with BSI from burn wounds compared with urinary or CRBSIs [23]. While there are conflicting data concerning the contribution of inhalation injury, higher deaths in BSI have been associated consistently with larger TBSA of burn (>45%–50%; odds ratio [OR] 1.05) and recurrent bacteremia (OR 9.12–41.6) [8,9,14,23].

It remains unclear whether polymicrobial BSI is independently associated with death. In a single study, while patients had higher mortality rates with polymicrobial than monomicrobial BSI (32% vs. 15%), those with polymicrobial BSI had more severe burns, raising the question of whether this was more a reflection of the severity of illness [23].

In conclusion, with advances in the treatment of burn patients, the epidemiology of BSI has evolved, with a greater focus on nosocomial, recurrent, and polymicrobial BSI.

Endocarditis

Endocarditis in the burn unit has been called the “silent source of sepsis” because of the difficulty in making the diagnosis. Based on two retrospective reviews from a large burn center from 1969 to 1974 and 2003 to 2006, the cumulative incidence of endocarditis in the burn population has been estimated to be 0.4%–1.3%, four times that of the general population [24,25]. Because of differing methodologies, it is difficult to make a direct comparison with other high-risk groups, but incidence and mortality rates may be higher in the burn patient (Table 2). Therefore, understanding the risk factors, presentation, microbiology, and unique aspects of endocarditis in the burn patient is crucial.

Burn patients with endocarditis often lack traditional risk factors, with the majority of cases involving structurally normal heart valves [24–27]. In a series by Regules et al. [25], only one of five patients had a structural predisposition, but this was associated with earlier onset of BSI and death. Theories on the involvement of structurally normal valves in this population range from a predisposition to forming fibrin-platelet vegetations with the transient burn-associated hypercoagulable state to recurrent endocardial trauma from indwelling catheters [28,29].

In a retrospective review of nosocomial endocarditis, 17.3% of patients had a previous indwelling line [30]. If endocarditis was related to recurrent line-related trauma, the majority of cases should be right-sided (as was supported by a 1970s case series in which 75% of endocarditis was right-sided or bilateral), but with transition from clinical or autopsy diagnosis to echocardiography, left-sided disease became prominent (71% of cases) [24,31].

The most common cause of endocarditis in the general, nosocomial, and burn population (almost half of described cases in burns) is S. aureus [25,29,31–40]. Enterococcus accounts for 10%–30% of general and nosocomial endocarditis (even higher risk with prosthetic valves), compared with 5% in the burn population [32,41]. There is a higher prevalence of gram-negative involvement in burn-associated endocarditis, with 17%–21% of burn endocarditis secondary to P. aeruginosa compared with 3% in the general population [25,31,42]. Polymicrobial infections account for 1%–6.8% of endocarditis in the general population compared with 36%–50% in burns [25,31,33,43].

The Duke criteria for diagnosis of infective endocarditis have an estimated sensitivity and specificity of 71%–99% and 99%, respectively, in the general population [44,45]. Duke criteria are difficult to apply to burn patients for a multitude of reasons, including, but not limited to the fact that burns and subsequent wound coverage may obscure vascular and immunologic phenomenon. Based on case series, only 11% of burn patients with a diagnosis of endocarditis fulfilled the Duke criteria for definite endocarditis [25]. Murmurs are rarely reported with burn endocarditis (at approximately 9%–28%), and it is difficult to tell from the literature whether these were pre-existing [24,25,31].

In published case series, the mean TBSA of adult burn patients with endocarditis was 5%–62% [25,31]. Persistent bacteremia is both a diagnostic criteria and risk factor for endocarditis [27,28,44]. In retrospective studies of burn endocarditis, endocarditis will develop in 2%–9% of patients with persistent BSI [24,25]. Therefore, endocarditis should be considered in burn patients with higher TBSA (>40%) and persistent BSI, with one previous study recommending early transesophageal echocardiography in all burn patients with persistent BSI without a known source [25]. This may be difficult to define in the burn population, however; line exchanges may be delayed because of limited access [46,47].

Another issue is the approach to valvular repair in burn patients because of challenges with traditional surgical approaches. For instance, the traditional median sternotomy site may be through a colonized or infected thoracic burn site. If grafting has already taken place over the desired site of the sternotomy, it is unclear what wound complications may be expected. Case reports describe two alternative approaches to traditional surgical approaches in burn patients. One patient with extensive thoracic burn wound infection and colonization had a successful mitral valve replacement using a right-sided thoracotomy [26]. Another patient who already had initial debridement of his burn wounds underwent removal of his allograft, which was then sterilized with chlorhexidine at the time of sternotomy, and had successful mitral valve replacement followed by placement of a new allograft [48]. As with many complicated infections in the BICU, collaboration between the burn unit, infectious disease, and cardiothoracic surgery teams is essential.

CNS Infections

The CNS complications in patients with burns have been attributed to many causes including metabolic encephalopathies, cerebrovascular accidents, but, less commonly, infections [49–51]. Assessing the mental status of critically ill burn patients poses a challenge given the use of sedation and narcotics for pain control and potential for metabolic encephalopathy. The incidence of CNS infection may be underestimated, because many patients may not have cerebrospinal fluid (CSF) sampled in the setting of bacteremia.

In hospitalized patients, the risk for development of nosocomial meningitis is rare without antecedent head trauma or previous infection [52]. In neurosurgical ICU patients, however, retrospective reviews note an approximate 3–10% incidence of CNS infections with predominately gram-negative organisms such as Pseudomonas and Acinetobacter spp. and methicillin resistant Staphylococcus aureus (MRSA) [53].

In the past, the true incidence of CNS infection in burn patients was largely unknown because little data existed [49,50]. An early post-mortem study investigating CNS complications in patients with severe burns in a large urban hospital over a 20-year period (1968–1988) reported 16% had CNS infection with 10% of cases having brain abscesses; however, this was not found in a more recent autopsy review in burn patients [49,51]. Scattered reports describe brain abscesses associated with head and facial burns, likely a result from the direct invasion of organisms or dissemination after deep burns of the scalp or face [51,54–56]. It has also been reported that most CNS infections result from a systemic source such as a concurrent BSI with the same organism that is subsequently isolated from the CNS [51].

In a large retrospective review of 1,964 admissions to a burn unit caring for military and civilian patients between 2003 and 2008, the overall incidence of CNS infection was extremely low at 0.1% (two of 1964 admissions) [50]. While past reports suggest that facial burns and head trauma may be associated with an increased risk of CNS infection, this was not found in this review, where 28.5% of patients had head or neck trauma and facial burns [50,51].

One hundred and twenty-five patients had surgical procedures of the head to either harvest the scalp or for debridement of scalp burns; nine had an invasive surgical procedure involving penetration of the skull (four craniectomies, five trephination). Of 12 patients with suspected CNS infection, only the two patients who underwent craniectomies had bacteria isolated from their CSF. Both of these patients were soldiers with head trauma and facial burns who required intracranial operation. In addition, both had bacteremia with A. baumannii and MRSA, the same organisms isolated from their CSF. Scalp harvesting donor sites did not pose an increased risk for the development of CNS infections. Overall, the rate of CNS infections was low even in patients with head and face burns and trauma unless the patient underwent an intracranial procedure [50].

Rare Fungal Infections in the Burn Patient

Other sections of this supplement discussed the common invasive fungal infections (IFI) in burn patients. Similar to the molds described in the IFI section, the majority of these fungi are ubiquitous in nature. Because the majority of these IFI are from case reports or series, it is difficult to describe their overall incidence, risk factors, and treatment (Table 3). They are important to discuss, however, because they may be related directly to thermal injury, potentially affect death, and, in the case of outbreaks, have devastating effects on multiple patients. Discussed here are rare fungal infections, fungal outbreaks, and disseminated fungal infections.

Of the rare fungal infections, Trichosporon has the most literature reported in burn patients. Trichosporon fungal wound infections (FWI) are associated with a high TBSA (42%–85%). Cases of Trichosporon FWI, fungemia, meningitis, and cerebral abscess have been reported. The reported mortality rate is 75%, which may be from initial misdiagnosis because of this organism's rarity and morphologic similarity to Candida, resulting in inactive empiric anti = fungal use [56–58].

Other rare yeasts infecting burn patients are Saccharomyces spp. and C. auris. S. cerevisiae is a yeast used by the food industry and may be part of the human gastrointestinal microbiome. S. boulardii is common in commercially available probiotics. Both have caused fungemia in burn patients (one related to a biopsied esophageal ulcer and the other probiotics) [59–61]. While S. boulardii fungemia after probiotic exposure has been reported repeatedly in critically ill patients, there are limited data to guide antifungal choice [59]. Both burn patients reported were treated with amphotericin B and survived [60,61].

The emerging multi-drug–resistant (MDR) yeast, C. auris, has already been reported in a burn patient. A case report describes a South African patient transferred to a United States burn center with C. auris fungal wound colonization (FWC) that progressed to FWI. The literature is replete with challenging outbreaks in the BICU, and this emerging MDR organism's outbreak potential cannot be overemphasized [62].

Other rare molds that have been reported in burn patients are Lomentospora (formerly Scedosporium) and Curvularia, a dark mold. Lomentospora represents another MDR fungi that is intrinsically resistant to most antifungal agents [63]. There have been three cases of Lomentospora in burn patients reported [64,65]. In the first two cases, Lomentospora isolates were pulmonary at 21–30 days post-burn from patients with TBSA involvement of 40%–65%. Both patients were treated with voriconazole, and one survived, although determining colonization versus infection in this setting is challenging [64,66]. The last case of Lomentospora was a FWI in a patient with 43% TBSA burns. His wounds were initially grossly contaminated with mud, with concern for infection on day seven of hospitalization followed by death within 24 h [65].

Curvularia has also complicated grossly contaminated burn wounds early in hospitalizations. Two patients described had TBSA burns of 50%–60% and lesions by hospital day four had immediate debridement (plus amphotericin B for the FWI) and survived [67].

In burn units, there have been outbreaks of Aspergillus, mucormycosis, Fusarium, and Trichosporon [56,68–72]. While early FWI may be related to exposures at time of injury, multiple patients with FWIs with the same organism more than two weeks into hospitalization should trigger concern for an outbreak. An outbreak investigation, involving infection prevention and control and environmental services, should first create a case definition and compare to baseline rates. For an outbreak, the cases are then reviewed for commonalities and potential areas for testing. In immunocompromised patients, outbreaks of fungus have also been linked to hospital linens, medical equipment, and adjacent construction [55,72–75]. Because some fungi form spores that can aerosolize, filters in the patient rooms in addition to medical equipment should be investigated [70,72].

These investigations of environmental samples can be exceedingly difficult and require multi-disciplinary approaches [76,77]. Even when a thorough outbreak investigation occurs, the source of an outbreak is not always identified. One burn center noted four patients with wound cultures with Fusarium and shared operating room exposure. While all environmental and air quality samples were negative, with improvement of terminal cleaning practices in the operating and patient rooms, the outbreak resolved [71].

The angioinvasiveness of some fungi allows dissemination, which is associated with an attributable mortality rate of approximately 30% [74,78–80]. A 10-year retrospective review of burn patients approximated that 21% of FWIs disseminated, with the greatest risk factor being TBSA >40% and the majority from pulmonary sources [81]. Disseminated fungal disease often goes unrecognized by providers, with ante-mortem diagnosis made only in 15%–40% of cases [82].

One of the diagnostic challenges is the difficulty in culturing many fungal species. While Candida, Trichosporon, Aspergillus, and Fusarium species can grow in bacterial blood culture vials, the yield and significance of these blood cultures is highly variable [56,83,84]. A retrospective review of Fusarium sp. in a pediatric burn unit found that 50%–70% of the disseminated Fusarium cases had positive blood cultures [83]. In studies of non-burn patients with invasive aspergillosis, however, only 10% had positive blood cultures, and only 5% of blood cultures with Aspergillus had evidence of invasive disease, leading to low negative and positive predictive value [84].

Many adjuncts to diagnosing mold infections have been found to be low yield in the burn population. In a retrospective study of burn patients with a median TBSA 29%, no correlation between FWC or FWI and 1-3 β-glucan was found (positive and negative predictive value 27% and 0%, respectively). Unfortunately, there were not enough Aspergillus FWI to determine the predictive value of galactomannan [85].

While the main treatment of FWI is debridement, systemic antifungal therapy is used as an adjunct. Given the varying susceptibility of molds, empiric coverage with amphotericin and a broad-spectrum triazole is recommended (Table 3). There are growing concerns for increasing antifungal resistance because of inappropriate use of antifungal agents, fungicides in agriculture, and antibiotic agents [39,86–88]. Death from these infections is high, and a high index of suspicion is essential to start early surgical debridement and systemic antifungal therapy.