A, A box plot showing the date of surgery vs the duration of prophylactic antibiotics. Prior December 2009, all patients received prolonged courses of prophylactic antibiotics, although the length was highly variable (median, 10.0 days). Starting in December 2009 we began to use shorter (≤24 hour) courses of antibiotics in many patients, although courses of longer duration were still highly variable (median, 1.5 days) (P < .001). Boxes represent interquartile ranges (IQRs) of the data, while each whisker represents data within 1.5 IQR of the nearest quartile. Outliers are shows as individual points. B, A histogram showing the duration of prophylactic antibiotics for each patient, demonstrating that 1-day and 7-day courses were the most commonly used.
Mitchell RM, Mendez E, Schmitt NC, Bhrany AD, Futran ND. Antibiotic Prophylaxis in Patients Undergoing Head and Neck Free Flap Reconstruction. JAMA Otolaryngol Head Neck Surg. 2015;141(12):1096–1103. doi:10.1001/jamaoto.2015.0513
Evidence supports short courses of perioperative antibiotics for patients receiving minor head and neck procedures. Few studies have addressed antibiotic prophylaxis for patients undergoing free flap reconstruction of head and neck defects.
To determine ideal antibiotic prophylaxis in patients undergoing head and neck free flap reconstruction.
Design, Setting, and Participants
Retrospective cohort study of 427 adults receiving free flap reconstruction of head and neck defects at 2 affiliated tertiary care academic hospitals between January 1, 2006, and January 28, 2013.
Prophylactic antibiotic type and duration were recorded from patient records.
Main Outcomes and Measures
Outcome data were abstracted from patients’ medical records including infection at the surgical sites and distant nonsurgical sites and flap site complications including flap compromise, dehiscence, or fistula. Multivariate logistic regression was used to determine the association of risk factors with the primary outcome of any infection within 30 days of surgery.
Ninety-six patients (22.5%) received prophylactic antibiotics for 24 hours or less, and 331 patients received prolonged courses of prophylactic antibiotics. The majority of patients received ampicillin-sulbactam alone for prophylaxis (53.2%), while 36.5% received clindamycin alone and 10.3% received an alternative regimen. Postoperative infections occurred in 46% of patients, and 22% of patients had an infection at the flap inset site or neck incision. The use of clindamycin (odds ratio [OR], 2.54; 95% CI, 1.25-5.14 [P = .01]) was associated with an increased risk of postoperative infection; extended duration of antibiotics (OR, 0.63; 95% CI, 0.34-1.19 [P = .18]) was not associated with increased risk of postoperative infection. By multivariate analysis, use of clindamycin (OR, 6.71; 95% CI, 1.83-24.60 [P = .004]) and oral tobacco use (OR, 1.20; 95% CI, 1.04-1.39 [P = .02]), but not extended course of prophylactic antibiotics (OR, 0.75; 95% CI, 0.30-1.86 [P = .53]), were associated with a higher risk of postoperative flap or neck infections.
Conclusions and Relevance
The choice of antibiotic appears to affect the rate of all postoperative infections and flap site infections more than the duration of antibiotics following head and neck free flap reconstruction. At our institutions, ampicillin-sulbactam is the preferred prophylactic antibiotic for major clean-contaminated head and neck procedures when possible.
Prevention of infections after procedures involving the head and neck has been the subject of extensive debate and a number of studies. Many procedures in the head and neck involve a breach of mucosal surfaces, and bacterial contamination of surgical wounds may lead to poor outcomes. These procedures are associated with a high rate of surgical site infections without the use of prophylactic antibiotics.1 Evidence generally supports using antibiotic prophylaxis in patients undergoing major head and neck clean-contaminated procedures.1,2
Quiz Ref IDGuidelines published by the American Society of Health-System Pharmacists, in conjunction with the Infectious Disease Society of America, Surgical Infection Society, and Society for Healthcare Epidemiology of America,3 recommend consideration of a single dose of prophylactic antibiotic only in clean procedures when prosthetic materials are placed. They generally recommend 24 hours of antibiotic prophylaxis for most major clean-contaminated procedures. However, there have been relatively few studies that specifically address antibiotic prophylaxis in patients receiving free flap reconstruction in the head and neck. Many patients receiving free flaps have a number of characteristics considered possible risk factors for developing postoperative infections and other complications, including the use of tobacco and alcohol; long procedure duration; previous resection, radiation and chemotherapy; poor nutritional status; and multiple comorbidities.3,4
This study was performed to assess the risk of postoperative infection associated with the duration of prophylactic antibiotics and other potential risk factors in patients receiving free flap reconstruction of defects involving the head and neck. A change in the recommendations for antibiotic prophylaxis at our institutions allowed us the opportunity to examine the association of antibiotic duration with postoperative outcomes. We hypothesized that short (≤24 hour) courses of prophylactic antibiotics would be associated with increased risk of postoperative infections in general and specifically at the flap inset site.
This study was designed as a retrospective cohort study. The institutional review board of the Human Subjects Division at the University of Washington approved this study. All patients 18 years or older who had free flap reconstruction of head and neck defects performed at the University of Washington Medical Center or Harborview Medical Center, Seattle, Washington, between January 1, 2006, and January 28, 2013, were identified.
Patients’ electronic medical records were examined to determine surgical wound classification. We excluded patients whose surgical wounds were considered contaminated or dirty as classified by preoperative assessment or by pathology reports indicating current infection. If patients received a revision flap within 30 days of the initial reconstruction, we excluded the second procedure from the analysis. Data were abstracted by a single author from patients’ medical records including date of surgery, duration of perioperative antibiotic prophylaxis, type of antibiotic, methicillin-resistant Staphylococcus aureus (MRSA) carrier screen test results, age, sex, free flap donor site and flap components, surgical wound classification, pathologic condition resulting in surgical defect, cervical lymph node stage (N stage), age-adjusted Charlson comorbidity index (CCI), diagnosis of hypothyroidism, previous resection, previous radiation therapy to the head and neck, previous chemotherapy, alcohol use at the time of the preoperative appointment, tobacco smoking history, and oral tobacco use. Hypothyroidism was defined as an elevated thyrotropin (TSH) level (>5.0 mIU/L).
Clinical outcomes within 30 days of the reconstruction were determined from patients’ medical records including infection at the flap site, neck incision, tissue donor site, and distant nonsurgical sites (eg, pneumonia, urinary tract) and flap site complications including dehiscence, fistula, or flap compromise. Infection was determined by interpretation of clinical records by a single author (R.M.M). Criteria in the medical records that indicated infection included use of the words “infection,” “cellulitis,” “pneumonia,” “UTI,” or “bacteremia” or other similar wording; positive culture results; or imaging combined with clinical evaluation (eg, pneumonia). Clinical signs of infection included purulent drainage, significant and worsening edema and/or erythema, and the clinician’s decision to continue or restart antibiotic therapy. For the purposes of this study, dehiscence was defined as any observable opening at the flap suture line, and flap compromise was defined as any tissue loss or debridement other than superficial epidermolysis or vascular insufficiency requiring revision. Our primary end point was any infection occurring within 30 days of the reconstruction. Table 1 gives the descriptive statistics of the independent variables and outcomes of the included patients.
Duration of antibiotics was compared using the Mann-Whitney test between 2 groups: those receiving free flap reconstruction prior to December 2009, when we began using shortened courses of prophylactic antibiotics, and those receiving free flap reconstruction after this time. Univariate analysis comparing outcomes was performed with logistic regression. Each multivariate analysis was performed using logistic regression incorporating all independent variables. Differences were considered statistically significant at P < .05. Statistical analyses and figures were generated using STATA, release 12.0 (StataCorp).
A total of 516 head and neck free flap reconstructions were performed in adults by our department at 2 hospitals during the study period. Eighty-three flaps were excluded because the wounds were considered contaminated or dirty. Six patients received 2 flaps within 30 days as a revision, and the second of each of these was excluded from analysis. Quiz Ref IDIn total, 427 clean and clean-contaminated procedures were included in the final analysis. These 2 surgical wound classifications were grouped together because they typically receive the same antibiotic regimen. We began screening patients for MRSA by nasal swab on admission in January 2009, and 239 of the patients included in this study had screening performed.
During the time course of our study, hospital-wide and surgeon preferences have changed regarding the duration of perioperative antibiotics given to patients with clean and clean-contaminated surgical wounds. Prior to December 2009, all patients in our study received prolonged courses of perioperative antibiotics, although the overall length was variable. Between 2009 and 2011, there was a transition to shortened (≤24 hours) courses of antibiotics in some patients. Since 2011, nearly all patients receiving free flaps have received prophylactic antibiotics for 24 hours or less. During the study period, antibiotic prophylaxis was continued for 1 week or until all drains were removed, hospital discharge, first follow-up appointment, or various other time points. If a note was placed in the patient’s medical chart indicating that antibiotic use was continued because of a concern about infection, this was recorded as such in the patient’s outcome. Figure, A, shows a comparison of the duration of antibiotic prophylaxis in patients treated prior to December 2009 vs those treated after this time. Patients treated prior to December had significantly longer durations of antibiotic prophylaxis compared with those treated in the later period (median, 10.0 days vs 1.5 days; P < .001). Figure, B, shows a histogram of the duration of antibiotics prophylaxis for each patient, demonstrating that 1-day and 7-day courses of antibiotics were the most commonly used.
Hospital-wide policies at our institutions have intermittently favored either ampicillin-sulbactam or clindamycin for antibiotic prophylaxis for clean-contaminated head and neck procedures. Ultimately, the choice and duration of the antibiotic were decided by the operating surgeon. For the patients included in the study, 96 (22.5%) received prophylactic antibiotics for 24 hours or less, while 331 patients received prolonged courses of prophylactic antibiotics. The majority of patients received ampicillin-sulbactam alone for prophylaxis (n=227 [53.2%]), while 156 (36.5%) received clindamycin alone and 44 (10.3%) received an alternative regimen. Because of the small sample size in the latter group, these patients were grouped together for analysis.
The choice of flap donor site depended on the specific tissue defect, patient anatomy and comorbidities, and surgeon preference. Free flap choices included the following: radial forearm, 148 (34.7%); fibula, 133 (31.1%); anterolateral thigh, 68 (15.9%); latissimus dorsi, 20 (4.7%); iliac crest, 16 (3.7%); jejunum, 16 (3.7%); rectus abdominis, 12 (2.8%); and scapula, 8 (1.9%). We performed 5 anterolateral thigh/fibula free flap simultaneous combinations (1.2%), and 1 latissimus dorsi/fibula free flap simultaneous combination (0.2%). For our analysis, patients were grouped according to whether the flap consisted of all soft tissue (252 flaps [59.0%]) or bone and soft tissue (175 flaps [41.0%]).
Quiz Ref IDOverall, 195 patients (45.7) had at least 1 infection occurring at either a surgical site or distant site in the postoperative period. Flap site or neck infections occurred in 93 patients (21.8%), while donor site infections occurred in 52 patients (12.2%) and distant infections occurred in 73 patients (17.1%) (Table 1).
Table 2 gives the association of each independent variable with the primary end point (ie, any infection) by univariate analysis. A postoperative infection was diagnosed in 55 patients (57%) receiving 24 hours or less of antibiotic prophylaxis and in 141 patients (42%) receiving prolonged courses of prophylactic antibiotics. Use of clindamycin (odds ratio [OR], 2.58; 95% CI, 1.70-3.93[ P < .001]), hypothyroidism (OR, 6.36; 95% CI, 1.81-22.37 [P < .001]), and osteocutaneous flaps (OR, 1.48; 95% CI, 1.01-2.18 [P = .05]) were significantly associated with increased risk of postoperative infection, while extended duration of antibiotic prophylaxis (OR, 0.55; 95% CI, 0.34-0.87 [P = .01]) was significantly associated with reduced risk of postoperative infection. Table 3 gives the association of each independent variable with the primary end point by multivariate analysis using logistic regression with each independent variable included in the model. Quiz Ref IDAfter correcting for all independent variables, only the use of clindamycin (OR, 2.54; 95% CI, 1.25-5.14 [P = .01]) and hypothyroidism (OR, 10.79; 95% CI, 2.76-42.24 [P = .001]), but not extended duration of antibiotics (OR, 0.63; 95% CI, 0.34-1.19 [P = .18]), were associated with an increased risk of postoperative infection. Including the variable of reconstructive surgery prior to or after December 2009 in the multivariate model did not significantly alter the results (data not shown).
Because the primary goal of perioperative prophylactic antibiotics, arguably, is a reduction of infections at the resection and flap site, we performed subgroup analysis to identify risk factors associated with infection at the flap site or neck, as shown in Table 4. In a multivariate model incorporating all independent variables, use of clindamycin (OR, 5.62; 95% CI, 2.10-15.06 [P = .001]) and hypothyroidism (OR, 5.26; 95% CI, 1.35-20.60 [P = .02]) were associated with an increased risk of flap site or neck infections. In this model, extended duration of prophylactic antibiotics (OR, 0.45; 95% CI, 0.21-0.98 [P = .04]) and clean-contaminated surgical wounds (OR, 0.26; 95% CI, 0.07-0.98 [P = .05]) were associated with decreased risk of flap site or neck infections.
Infections are thought to contribute to further flap site complications including fistula formation, flap compromise, or flap loss. We performed subgroup analysis to identify risk factors associated with all flap site complications. In a multivariate model incorporating all independent variables, use of clindamycin (OR, 2.77; 95% CI, 1.35-5.66 [P = .01]), hypothyroidism (OR, 4.14; 95% CI, 1.27-13.50 [P = .02]), and osteocutaneous flaps (OR, 2.29; 95% CI, 1.23-4.26 [P = .01]) were associated with increased risk of any flap site complications. By multivariate analysis, compared with soft-tissue flaps, osteocutaneous flaps were significantly associated with increased risk of flap dehiscence (32.6% vs 19.8%; OR, 2.47; 95% CI, 1.25-4.88 [P = .01]) and flap compromise (14.9% vs 6.7%; OR, 4.39; 95% CI, 1.45-13.27 [P = .01]). There was a greater risk of fistula (12.6% vs 7.1%; OR, 2.40; 95% CI, 0.96-6.04), which was not significant at P = .06.
There were a large number of patients receiving antibiotics for 7 days (n = 40) and 8 days (n = 58). We performed an analysis that included only patients receiving antibiotics for 24 hours or less and those receiving antibiotics for 7 to 8 days. When incorporating all variables into the multivariate model examining factors associated with the primary end point (ie, any infection), the use of clindamycin (OR, 4.95; 95% CI, 1.47-16.47 [P = .01]) and osteocutaneous flaps (OR, 3.04; 95% CI, 1.15-8.01 [P = .02]), but not prolonged duration (OR, 0.40; 95% CI, 0.13-1.24 [P = .12]), were associated with higher risk of infection. When incorporating all variables into the multivariate model examining factors associated with the secondary end point (ie, flap or neck infections), only the use of clindamycin (OR, 3.94; 95% CI, 1.51-10.29 [P = .01]) was associated with higher risk of infection.
Because of the high rate of infections in clean surgical wounds, we performed subgroup analysis of clean-contaminated surgical wounds only. In a multivariate model incorporating all independent variables, use of clindamycin (OR, 6.71; 95% CI, 1.83-24.60 [P = .004]) and oral tobacco use (OR, 1.20; 95% CI, 1.04-1.39 [P = .02]), but not extended course of antibiotic prophylaxis (OR, 0.75; 95% CI, 0.30-1.86 [P = .53]), were associated with a higher risk of postoperative flap or neck infections.
Overall, our results suggest that the use of clindamycin for perioperative antibiotic prophylaxis and hypothyroidism was associated with increased risk of overall postoperative infections, infections at the flap site or neck, and overall flap site complications. Prolonged courses of perioperative antibiotics were associated with a decreased risk of infections at the flap site or neck, but the association was not significant when considering all postoperative infections together or just the clean-contaminated cases. Clean-contaminated wounds were associated with lower rates of infection at the flap site or neck, and a history of oral tobacco use was associated with increased risk of flap site infections in patients with clean-contaminated wounds, while osteocutaneous flaps were associated with a greater risk of flap site complications but not infections. By multivariate analysis, none of the other potential risk factors we included were significantly associated with these clinical outcomes.
We hypothesized that the duration of perioperative antibiotics would be associated with the risk of postoperative infection. In patients with head and neck cancer not receiving free flap reconstruction, a number of studies have supported the use of short-term prophylactic antibiotics.5- 7 Our hypothesis was based on the assumptions that patients receiving free flaps would have more advanced disease, more drains placed and for longer periods, and longer procedure times compared with patients not receiving free flaps. A previous prospective study compared 3 vs 15 doses of clindamycin in 74 patients with head and neck cancer receiving free flaps with clean-contaminated surgical wounds.8 Although no differences in any infection or any other outcome were significantly different between the 2 treatment groups, this study was underpowered to detect differences based on the low rate of infections. In addition, infection was only assessed within 7 days of the procedure or until the patient was discharged from the hospital. Longer follow-up of these patients may have detected more infections. Another study found the average time for diagnosis of a postoperative infection was 30 days after head and neck reconstructive surgery.9 With longer follow-up and potentially increased rates of infection, the power to assess differences between the treatment groups may have been greater, although possibly insufficient given the small sample sizes.
We observed that the use of clindamycin was independently associated with higher risk of overall infections, flap/neck infections, and flap site complications. Although no cultures were obtained from the majority of these infections, we hypothesize that this may be related to poor coverage of aerobic gram-negative bacteria. Others have shown that gram-negative organisms are more commonly cultured from infected sites in head and neck surgery patients receiving clindamycin vs those receiving ampicillin-sulbactam.10 This is a limitation of clindamycin, and it is sometimes combined with another antibiotic when gram-negative bacteria are targeted.2,11
Undertreated hypothyroidism was significantly associated with increased risk of overall and flap site postoperative infections and flap site complications in our study. Hypothyroidism is relatively common in patients with head and neck cancer, particularly those who have underwent prior resection or radiation therapy.4 Many of our patients did not have thyroid function testing performed, and there was a tendency to test only after the development of complications. This likely skewed our results to overemphasize the effect of hypothyroidism. However, hypothyroidism has a plausible association with postoperative infection and complications related to wound healing,4 and hypothyroidism has been shown to impair blood clotting,12 which could lead to increased risk of hematoma formation. Further study of the association of hypothyroidism with outcomes in flap patients is warranted to elucidate the underlying mechanisms.
Osteocutaneous flaps were associated with higher rates of flap dehiscence, flap compromise, and flap site complications as a group. In contrast to soft-tissue flaps, osteocutaneous flaps typically require placement of hardware, which can facilitate bacterial colonization. Although infections at the flap site would be expected to increase the risk of these flap site complications,13 the greater complexity of defects requiring osteocutaneous flaps, and the greater demand on tissue healing associated with these flaps might explain the greater number of complications we observed.
Clean-contaminated surgical wounds were associated with lower risk of flap/neck infections in our study, which is in contrast to previous studies.14 Of 42 surgical wounds classified as clean, there were 8 infections involving the flap site or neck. Of the patients who developed an infection, 3 received ampicillin-sulbactam, 3 received clindamycin, and 2 received cefazolin. Only 2 patients had positive culture results, and so it is possible that infections were overdiagnosed. Although statistically significant, the low power resulting from the small number of patients with clean surgical wounds and the small number of infections within this group reduces the possibility that this represents a true effect. A study powered to detect the effect of surgical wound classification on postoperative infection is necessary to further elucidate this possible relationship.
Quiz Ref IDA broad number of risk factors including sex, alcohol use, smoking, previous resection or radiation, diabetes mellitus, and age-adjusted comorbidity score have been proposed to affect the rate of infection and postoperative complications in patients receiving free flap reconstruction of head and neck defects.9,15 Previous studies have shown variable results with many of these potential risk factors,9,14- 16 which is likely due to the wide variety of preoperative comorbidities and previous treatments, surgical indications, patient selection, and surgeon preferences for postoperative treatment. Research methodology and data analysis is also highly variable among studies.8,9,15 We believe that multivariate analysis is critical to account for the large number of potential confounding variables when studying this patient population.
Limitations of this study include those inherent to retrospective studies and comparatively few patients receiving short courses of prophylactic antibiotics. Patients receiving prolonged courses of antibiotics had highly variable lengths of prophylactic treatment. Although this variety is a potential limiting factor in grouping all of these patients together, the variable courses of antibiotics for the patients included in this study are not due to concern for infection. Any antibiotic treatments that were continued owing to concern for infection were recorded as a positive infection outcome. It is possible that some of the variety in the length of prophylactic antibiotic treatment was due to subjective concern for postoperative infection that was not indicated in the medical record. In addition, the primary outcome in this study was infection at the surgical sites or in distant locations. The diagnosis of infection frequently relied on clinical judgment, and culture results were not possible in the majority of cases. Many patients had topical antibiotic ointment applied to external incisions, and some of the suspected infections at these sites may have actually represented local tissue reactions to components of the ointment. The use of multiple antibiotics allowed us to also analyze the association of antibiotic choice with our outcomes, but also limited the power of this study to determine the effect of duration of antibiotic use. Future studies may examine these 2 separate factors in isolation to better determine their association with postoperative infection risk.
The use of clindamycin was significantly associated with greater risk of postoperative infection and flap site complications. As a secondary end point, prolonged duration of antibiotic use was associated with reduced risk of infection at the flap or neck sites when examining both clean and clean-contaminated cases, but not for clean-contaminated cases alone, which included most of the patients in our study. Future studies are warranted to determine whether these findings would justify prolonged therapy and the optimal duration of therapy. At our institutions, we have begun to use ampicillin-sulbactam as the preferred prophylactic antibiotic. Clindamycin is the second-line agent of choice for patients with a penicillin allergy, and based on results from this study we recommend the addition of a second antibiotic to improve coverage of aerobic gram-negative bacteria. These data support the additional study of these risk factors in patients receiving free flap reconstruction of head and neck defects in prospective trials.
Submitted for Publication: November 19, 2014; final revision received February 5, 2015; accepted February 24, 2015.
Corresponding Author: Neal D. Futran, DMD, MD, Department of Otolaryngology–Head and Neck Surgery, University of Washington, 1959 NE Pacific St, PO Box 356515, Seattle, WA 98195 (firstname.lastname@example.org).
Published Online: April 23, 2015. doi:10.1001/jamaoto.2015.0513.
Author Contributions: Dr Mitchell had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Mitchell, Mendez, Schmitt, Futran.
Drafting of the manuscript: Mitchell, Mendez, Futran.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Mitchell, Mendez.
Administrative, technical, or material support: Mitchell, Mendez, Futran.
Study supervision: Mendez, Schmitt, Bhrany, Futran.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This study was presented as a poster at the Annual Meeting of the American Head and Neck Society; April 21-22, 2015; Boston, Massachusetts.