We conducted a retrospective analysis of our trauma registry to identify all patients presenting to our level-I trauma center with open fractures who were subsequently treated with use of a standardized protocol for the management of open fracture wounds. Four hundred and twenty-two open fractures were treated operatively between March 2003 and December 2005. The inclusion criteria for this study were an open fracture of any type involving any extremity, for which the entire fracture management was under the care of our orthopaedic trauma service. All clinical information was collected from inpatient and outpatient medical records, and the study was approved by our institutional review board.
All patients were initially evaluated in the emergency department. At that initial evaluation, all open wounds were irrigated with normal saline solution and dressed with sterile gauze and the limb was immobilized in an appropriate fashion to facilitate further patient management. The patients were given tetanus and antibiotic prophylaxis, as dictated by the initial assessment of severity and contamination of the open fracture. All patients with an open fracture received intravenous cephalosporin and aminoglycoside antibiotics in the emergency department. If the wounds were severely soiled or contaminated, a penicillin derivative was also given. If the patient was allergic to penicillin, vancomycin was substituted for the cephalosporin.
Following clearance by the general surgery trauma service, the patients deemed stable for general anesthesia were taken to the operating room for further evaluation, thorough wound irrigation and debridement, and stabilization of the fracture with use of temporizing or definitive techniques. In most cases, pulsatile lavage was utilized with standard volumes of irrigant: Type-I fractures received 3 L of normal saline solution; Type-II fractures, 6 L; and Type-III fractures, 9 L. No irrigant additives were used. Following irrigation and debridement, aerobic and anaerobic cultures were obtained by swabbing the bone and local tissue directly at the site of the fracture only. Usually, Type-I wounds were packed open lightly with saline solution-soaked fine mesh gauze or loosely closed with nonabsorbable suture. The decision to leave the Type-I wound open or closed was at the discretion of the treating surgeon and was based on the apparent wound health and the degree of soft-tissue damage. Most Type-II and III wounds were left open and sealed with a negative-pressure vacuum-assisted closure sponge (V.A.C.; KCI, San Antonio, Texas), and an appropriate splint was applied. Thirty-two selected Type-II wounds were closed loosely at the initial debridement if the treating surgeon thought that there was minimal soft-tissue injury and no contamination. Antibiotic bead pouches were not utilized in any of the open fractures.
The patient was subsequently returned to the operating room at forty-eight hours for repeat irrigation and debridement of the open wound and wound closure if the cultures obtained after irrigation and debridement were negative. Seventy-three patients with low-energy injuries, negative original cultures, and well-approximated wounds were not returned to the operating room for secondary closure. Their wounds were allowed to close by secondary intention.
If the original cultures obtained after irrigation and debridement were positive for organisms, the wound underwent an additional irrigation and debridement procedure; additional cultures following irrigation and debridement were obtained, and the negative pressure dressing was reapplied. This process was repeated every forty-eight hours until the cultures were negative. The wound was then closed primarily or an appropriate soft-tissue coverage procedure was performed. In some cases, the cultures would initially show no growth at forty-eight hours but would turn positive at a later date, well after the wounds had been closed. These patients were reevaluated clinically and followed closely. If the wound appeared benign, it was not treated with repeat debridement; rather, the patient was continued on antibiotics specific to the organism in the latest positive culture. The antibiotic was continued until it was determined that the wound was healed without drainage. The average duration of treatment with antibiotics was three weeks, with some antibiotics being continued for up to six weeks depending on the cultured organism.
During open wound management, if positive cultures were obtained, the patient remained on an intravenous antibiotic specific for the cultured organism(s). The antibiotics were routinely discontinued twenty-four hours after wound closure unless late culture conversion occurred.
Following discharge from the hospital, the patients returned for follow-up at routine intervals. Clinical evaluation, including appropriate radiographs and wound inspection, was performed at each clinic visit. All patients were followed until they achieved fracture union. However, many patients required additional procedures in order to achieve fracture-healing if they progressed to delayed or obvious nonunion related to the fracture morphology itself and not to any occult deep infection. At the time of the secondary procedure required to achieve definitive fracture-healing, data collection was halted, assuming no signs of deep infection were present.
Statistical Analysis
Descriptive statistics of the sample characteristics were performed. The variables were analyzed with use of chi-square statistics for dichotomous comparisons and Mann-Whitney and Kruskal-Wallis nonparametric statistics for ordinal level variables. The alpha level for significance was set at 0.05. The predictor variable of delayed deep infection was analyzed with use of a logistic regression, with no deep infection coded as 0 and delayed deep infection coded as 1. We assessed the following variables: body mass index, Gustilo type of open fracture, age (<30 or =30 years), the presence of diabetes mellitus, and a history of smoking. When logistic regression was performed with use of upper extremity injuries as the subset, no variables were found to be significant; however, with the same analysis for lower-extremity injuries, Gustilo type was the only variable of significance. When upper and lower extremity injuries were combined, both Gustilo type and diabetes remained in the model for predicting deep infections.
Source of Funding
There was no external funding for this study.
Study Population
Four hundred and twenty-two fractures were evaluated during the study period; complete data sets were available for 346 open fractures, which made up the final study group. Twelve patients had more than one open fracture, and each fracture was evaluated individually. The mean age of the patients was 37.6 ± 15.2 years (range, fifteen to ninety-two years). There were 249 males (72%) and ninety-seven females (28%). Heights and weights at the time of admission were available for 332 patients, permitting the calculation of body mass index. The body mass index was <18.5 kg/m2 for five patients, between 18.5 kg/m2 and 24.9 kg/m2 for 116 patients, between 25 kg/m2 and 29.9 kg/m2 for 123 patients, and =30 kg/m2 for eighty-eight patients. The mean body mass index (and standard deviation) was 27.4 ± 6.1 kg/m2 (range, 16.6 to 71.6 kg/m2).
We used the Gustilo and Anderson system19 to grade the 346 fractures. There were fifty Type-I, 123 Type-II, 111 Type-IIIA, forty-seven Type-IIIB, and fifteen Type-IIIC fractures. With regard to irrigation and debridement procedures, Gustilo Type-I fractures underwent an average of 1.4 ± 0.6 procedures (range, one to four procedures); Type-II fractures, an average of 1.9 ± 0.7 procedures (range, one to six procedures); Type-IIIA fractures, an average of 2.4 ± 1.0 procedures (range, one to seven procedures); Type-IIIB fractures, an average of 4.9 ± 2.1 procedures (range, one to nine procedures); and Type-IIIC fractures, an average of 6.1 ± 2.5 procedures (range, three to ten procedures). Of the forty-seven Type-IIIB fractures, two involved amputations and forty involved soft-tissue coverage procedures (one with an amputation); the wound healed by secondary intention in six patients. Of the fifteen Type-IIIC fractures, nine were amputations and two were soft-tissue coverage procedures.
Infections
The overall rate of deep infection was 4.3%. No deep infection developed in the Gustilo Type-I fractures. Five (4%) of 123 Type-II fractures, and ten (5.7%) of 173 Type-III fractures had a deep infection. With the numbers studied, the increased risk for development of a deep infection in the Type-III fracture group was not significant (p = 0.187). A significant difference was detected among the three Type-III fracture categories (p = 0.003), with an infection found in two (1.8%) of 111 Type-IIIA fractures, five (10.6%) of forty-seven Type-IIIB fractures, and three (20%) of fifteen Type-IIIC fractures. The combined rate of deep infection for the sixty-two Type-IIIB and IIIC fractures was 12.9% (eight fractures) (Tables I and II).
Complete data on cultures and clinical wound follow-up were available for 346 fractures. A deep wound infection developed in fifteen patients, including two whose wounds were closed with positive cultures after debridement and thirteen whose wounds were closed with negative cultures after debridement. In the group of fractures without a subsequent deep infection, 80% were closed with negative cultures after debridement and 7% were closed with initially negative cultures that converted to positive later after closure had occurred. These were treated with culture-specific antibiotics. If the patient was still hospitalized, appropriate intravenous antibiotics were initiated. If the patient had been discharged, the appropriate oral antibiotics were begun. In forty-three (13%) of the 331 patients who had not developed an infection, the wounds were closed with presumably negative wound cultures, but these culture results could not be confirmed. In four of the fifteen patients with a deep infection, the infecting organism had been cultured previously during one of the interim debridements. The remaining eleven infections developed in wounds that had negative cultures after debridement throughout their care. The most common late infecting organism was a Staphylococcus aureus species (see Appendix).
The average number of days to closure of the wound, the average number of irrigation and debridement procedures, and the infection rates according to the location of the injury (upper or lower extremity) and Gustilo type are summarized in Tables I and II. The average duration of follow-up was 14 ± 12 months (range, six days to 4.5 years).
Group Comparisons
A history regarding additional comorbidities was available for 345 of the 346 fractures. Comorbidities included smoking (185 patients), diabetes mellitus (nineteen), illicit drug use (seventy-four), and alcohol use (204). When evaluated for categorical variables with use of chi-square analysis, the patients with diabetes mellitus demonstrated a significantly increased rate of deep infection compared with the study group as a whole, with a relative risk of 4.2 (p = 0.043). Patients with a body mass index of =25 kg/m2 had a significantly higher rate of deep infection than patients with a body mass index of <25 kg/m2 (6% and 1.6%, respectively; p = 0.047), and patients with a body mass index of =30 kg/m2 had a significantly higher rate than those with a body mass index of <30 kg/m2 (8% and 2%, respectively; p = 0.017). With the numbers studied, tobacco use did not appear to affect the infection rate, with a 5% infection rate in nonsmokers and a 3% infection rate in smokers (p = 0.588). With the numbers studied, we could not identify an association between time from injury to wound closure and infection.
There appeared to be no difference between upper and lower extremity injuries with regard to the rate of deep infection except that the rate of infection in Type-IIIB fractures was higher for injuries in the upper extremity (17.4%) than for those in the lower extremity (4.2%). However, the difference was not significant (p = 0.14). We believe this increase reflected the severe nature of the Type-III fractures in this specific subgroup and was not a result of the treatment protocol.
Significantly more irrigation and debridement procedures were done in fractures that developed a deep infection compared with those that had not (mean, 3.8 and 2.5 irrigation and debridement procedures, respectively) (p = 0.002). Twenty-four patients in the study group had the wounds closed despite the presence of positive wound cultures (a protocol breach), and two of them developed a subsequent deep infection. The risk of infection in these patients approached, but did not reach, significance (p = 0.0501).
The management of open fractures can be challenging. In this study, we attempted to define the role of cultures obtained after debridement in the management of the associated soft-tissue injury and as a guide to the timing of wound closure. Our protocol resulted in a very low overall rate of infection (4.3%) compared with several previous studies, in which overall rates were reported to have ranged from 7% to 22%20-25. Most dramatic was our very low rate of infection in all Type-III fractures (5.7%), particularly the Type-IIIA fractures (1.8%). These rates are much lower than the infection rates previously reported for these complex Type-III fractures, which have ranged from 10% to 50%19,20,26. Additionally, the overall low rate of infection in our study was much lower than the rates reported in series involving wounds that were closed primarily following an initial irrigation and debridement procedure, which have ranged from 7% to 20%3,4,27-29.
The occurrence of a positive culture after wound closure did not appear to have an effect on the rate of deep infection. However, the patients whose cultures became positive after closure were treated with culture-specific antibiotic therapy for up to six weeks. Antibiotic treatment is a confounding variable in the evaluation of whether late development of a positive culture after wound closure has a significant effect on the rate of deep infection. In two patients who developed a delayed infection, there were initial interval cultures that grew an organism that was identical to the organism causing the delayed infection. In the other thirteen delayed infections, no organisms grew on the interval cultures.
It is of note that seventy-three wounds were closed following the initial irrigation and debridement procedure. Of those, thirty-two were in a Type-I; thirty-two, in Type-II; eight, in Type-IIIA; and one was in a Type-IIIB fracture. Wound cultures were obtained following debridement in all of these fractures, and the wounds were closed with the intent to perform a repeat debridement if the cultures became positive. None of these patients developed a deep infection, and none had a positive culture after debridement. These were low-energy fractures with minimal soft-tissue and periosteal disruption, except for the size of the wound, which was used to classify the injury type. It is not our recommendation, however, that high-energy open fractures or fractures with extensive soft-tissue injuries be closed primarily. Primary wound closure should be considered only if the surgeon determines that the initial debridement has been thorough and the injury is a result of a low-energy mechanism.
It appears that the fractures most impacted by this protocol are the severe Type-III injuries. The impressively low rates of infection for this group in our study may have been related to the number of repeat debridements required to obtain a healthy, biologically competent soft-tissue envelope prior to definitive wound closure. This increased number of debridements comes at a higher cost to the health-care system as the use of this protocol may require multiple operative procedures and a prolonged length of stay. Furthermore, if the cultures turn positive after wound closure, the added expense of intravenous or oral antibiotic coverage is not insignificant.
We believe that a meticulous and thorough surgical debridement is the best defense against a postoperative infection. By following this principle, we believe that positive cultures obtained after debridement indicate an inadequate debridement, requiring a return to the operating room for additional debridement. Ultimately, in the treatment of open fractures, surgeon judgment should determine soft-tissue management. As a result of this study, we believe that cultures obtained after debridement can provide an objective measure to guide the timing of wound closure or coverage.