Surgical site infections are a major cause of perioperative morbidity and, as such, contribute to increased hospital costs.1 Numerous studies have demonstrated that perioperative antibiotic administration reduces infectious morbidity for high-risk surgical procedures. Along with bundled interventions aimed at reducing surgical site infections, the administration of perioperative antibiotics is recognized as standard of care for a number of procedures.2 The American College of Obstetricians and Gynecologists recommends antimicrobial prophylaxis for all modes of hysterectomy, along with other procedures associated with high infectious risk. The most recent update of these guidelines in 2018 suggests consideration of perioperative antibiotic administration at time of laparotomy, however recommends against antibiotic use in cases of operative laparoscopy.3 Despite such guidance, one retrospective study of more than half a million cases from a nationwide database reported that antibiotics are given in as many as 40% of cases for which they are not recommended, and specifically in 65.8% of myomectomies.4 Myomectomy was the third most frequently cited procedure associated with antibiotic use despite lack of recommendations for this practice.4 Another retrospective study of 1,045 cases in a teaching hospital reported antibiotic use in 96.3% of myomectomies.5
It is estimated that 34,000 myomectomies are performed annually in the United States,3,6 yet there is a paucity of literature investigating the appropriateness of perioperative antibiotic use for these cases. As such, clear guidance regarding antibiotic stewardship has the potential to minimize inappropriate antibiotic use and accompanying sequelae, or alternatively to optimize perioperative outcomes by decreasing infectious morbidity.
The primary objective of this study is to identify patterns of antibiotic use in a cohort of women undergoing myomectomy and report associations between perioperative antibiotics and risk of surgical site infections.
This retrospective cohort study included all patients who underwent a myomectomy by any route at Brigham and Women's or Brigham and Women's Faulkner Hospital, both in Boston, Massachusetts, from 2009 to 2016. Exclusion criteria included myomectomy performed at the time of cesarean delivery, preoperative concern for malignancy or additional planned major surgical procedure for which antibiotics were indicated. The patient cohort was identified with use of the institutional research patient data registry by querying for all cases of “myomectomy” during the study period. This study was approved by the Partner's Research Management Institutional Review Board.
The following demographic data and operative characteristics were abstracted from the electronic medical record: patient age, race, body mass index (BMI, calculated as weight in kilograms divided by height in meters squared), parity, prior abdominal surgery, indication for surgery, route of procedure (hysteroscopic, abdominal, laparoscopic or robot-assisted), surgeon type (general gynecologist vs fellowship-trained subspecialist in minimally invasive gynecologic surgery, reproductive endocrinology and infertility, gynecologic oncology, or urogynecology), administration of perioperative antibiotics (verified using anesthesia records), class of antibiotic if given, number of myomas removed, weight of myomas on final pathologic exam, location(s) of myomas, estimated blood loss (per surgeon estimate in the operative note), total operative time from incision to closure, intraoperative complications (including bowel or genitourinary tract injury, estimated blood loss greater than 1,000 mL, vascular injury, conversion to laparotomy from a minimally invasive approach or conversion to hysterectomy), entrance into the endometrial cavity at time of abdominal, laparoscopic, or robotic myomectomy, use of adhesion barrier or hemostatic agents, performance of chromopertubation, length of hospital stay in days, final pathologic diagnosis, and postoperative complications including infectious outcomes (scaled using the Clavien-Dindo complication rating, assessed up to 8 weeks postsurgery, detailed in Table 1).7 Any return to the operating room was classified as a postoperative complication.
Regarding infectious outcomes, the Centers for Disease Control and Prevention (CDC) criteria was used to define surgical site infection as an infection that occurs within 30 days after the operative procedure and involves the surgical incision, organ or space with qualifying clinical characteristics.8 The diagnosis of a wound infection was confirmed by symptoms of localized erythema; induration; warmth; pain at the incision site; purulent wound drainage; positive fluid culture obtained from a surgical site; a surgical site requiring exploration (either bedside or operatively) in the setting of at least one clinical sign of infection; or a surgeon's diagnosis of infection. Surgical site infections were further classified as superficial (involving only the skin or subcutaneous tissue), deep (involving deep soft tissues of an incision including fascia and muscle), or organ or pelvic space (any part of the anatomy other than the incision which was manipulated during the operative procedure). Other postoperative infections were recorded only if treated with antibiotics or other intervention, including urinary tract infections, pneumonia, vaginal infections, and fevers of unclear etiologies. Minor surgical site infections or other postoperative infections were defined as requiring treatment with outpatient management and oral medications. Major infections were defined as all deep or organ space surgical site infections and any other infection that required readmission and intravenous (IV) antibiotics. If no postoperative visit was documented, patient was assigned to the no surgical site infection group.
The myomectomy cohort was subdivided into the group that received perioperative antibiotics and the group that did not. Patients who received any antibiotics before or during the case were included in the cohort of those who received perioperative antibiotics. Owing to concern for confounding by alternate indication for antibiotic administration, patients who underwent chromopertubation at the time of myomectomy (n=221) or who experienced conversion to hysterectomy (n=1) were excluded from further analysis. Univariate analyses of patient characteristics and surgical outcomes between treatment groups were performed using χ2 and Fisher exact tests for categorical variables and either t test or Wilcoxon rank sum test for continuous variables. Routes of myomectomy defined as hysteroscopic or vaginal were not included in the further regression analyses owing to low incidence of infectious outcomes in these cases. For the purposes of the regression analysis, a composite variable of “high risk factor” for infection was created; this included all patients with any of the following characteristics: having BMI of 30 or higher, having diabetes, or being a current smoker. Multivariable logistic regression analyses regarding risk of infectious outcomes by antibiotic use were performed, controlling for patient age, route of surgery, presence of high risk factors, any intraoperative complication, myoma weight, and entrance into the endometrial cavity. The variables for regression analyses were selected as possible confounders that affected the decision to administer antibiotics, complexity of the surgery or the likelihood of surgical site infection.
A matched cohort analysis was also performed to corroborate findings in an attempt to correct for the underlying differences between groups. The matching criteria included age (within 5 years); exact route of surgery; high risk factors, defined as BMI of 30 or higher, current smoking, or diabetes; myoma weight (within 50 g or 100 g), and, if not available, number of myomas (within two); entry into endometrial cavity; intraoperative complications; and surgeon type. If more than one match was generated, the case with the closest myoma weight and BMI were selected as the final match. Data were analyzed using SAS software, and P=.05 was considered the cutoff point for statistical significance.
A total of 1,433 patients underwent myomectomy at our institutions between 2009 and 2016. Two hundred surgeries performed at Brigham and Women's Faulkner Hospital from 2009 to 2012, before initiation of standardized electronic medical records, were excluded from analyses owing to inability to locate anesthesia records and confirm administration of perioperative antibiotic. Excluding the patients who underwent chromopertubation (n=221) or experienced conversion to hysterectomy (n=1), the final myomectomy cohort was comprised of 1,211 patients. The vast majority of the patients, 92.7%, received perioperative antibiotics at the time of surgery. Of the patients who did receive antibiotics, 95.6% received a beta-lactam drug, most commonly 1–3 grams cefazolin administered intravenously. There were eight cases without a documented postoperative visit that were by coded as not experiencing postoperative complication or infection.
Baseline characteristics of the two subgroups (the group that received and that group that did not receive antibiotics) were similar, overall, and are summarized in Table 2. The majority of cases were performed in a minimally invasive fashion (66.6% of myomectomies). Additionally, 95.9% of all cases were performed by fellowship-trained subspecialists, predominantly by minimally invasive gynecologic surgery, or reproductive endocrinology and infertility surgeons. Among the group that did not receive antibiotics, 64.8% underwent laparoscopic myomectomy and 29.6% underwent hysteroscopic myomectomy; only 1–2% of cases in the no-antibiotic group represented an abdominal, robotic or vaginal approach. The patients who received antibiotics were, on average, 1.5 years younger than those in the no antibiotics group (38.3±6.6 vs 39.8±8.6 years in the antibiotic and no-antibiotic groups, respectively, P=.046). There were no differences detected between the groups with regard to race, BMI, parity or surgical history of laparotomy; the majority of the patients in both groups were white with an average BMI of 27–28. The most frequent indications for surgery in both groups were pelvic pain or pressure or menorrhagia. There was a very low and nondifferential rate of current tobacco smoking and diabetes. On matched cohort analysis, all baseline differences between groups disappeared apart from the discrepancy in antibiotic use by type of surgeon (Table 3).
Perioperative outcomes by group are delineated in Table 4. The cases that received antibiotics were associated with longer median operative times (140 [103–185] vs 85.5 [47–127.5] minutes in the antibiotic and no-antibiotic group, respectively, P<.001); higher median estimated blood loss (137.5 [50–250] vs 50 [17.5–100] mL in the antibiotic and no-antibiotic group, respectively, P<.001); greater mean number of myomas removed intraoperatively (7.2±10.5 vs 2.4±2.9 in the antibiotic and no-antibiotic group, respectively, P<.001); greater median myoma weight on final pathology (255 [105–467] vs 52.9 [13–234] grams in the antibiotic and no-antibiotic group, respectively, P<.001); and longer median length of stay (1 [0–2] vs 0 [0-0] days in the antibiotic and no-antibiotic group, respectively, P<.001). There was a higher frequency of entry into the endometrial cavity in the cases that received antibiotics (30.1% vs 13.6%, effect size 2.21, 95% CI [1.30–3.76], P=.001). A hemostatic agent was used in 5–8% of cases in both groups. There were no cases of occult malignancy noted on final pathology. There were 27 intraoperative complications (2.2% of all cases), 21 of those being estimated blood loss greater than 1,000 mL; the frequency of intraoperative complication was similar irrespective of antibiotic use. There was no difference in the incidence of postoperative complications between groups. There was one case of a Clavien-Dindo grade IVb surgical complication after laparoscopic myomectomy during which antibiotics were not given. In addition to leiomyomas, the patient had deep infiltrative endometriosis that required entry into the vagina for removal of a nodule. She was readmitted on postoperative day 5 with sepsis of unknown etiology, and underwent reoperation that was complicated by enterotomy and postoperative sepsis that was presumed to be from an ascending pelvic infection. This patient ultimately underwent a total abdominal hysterectomy and surgical debridement before clinical improvement.
Regarding the primary outcome, there was a greater, and statistically significant, incidence of any surgical site infection in the group that did not receive perioperative antibiotics (2.9% vs 6.8% in the antibiotic and no-antibiotic groups, respectively, effect size 0.43, 95% CI [0.18–0.97], P=.04). No difference was noted when outcomes were further stratified into major or minor infection, or by type of surgical site infection, although there was a higher incidence of any infectious outcome in the no-antibiotic group (5.8% vs 11.4% in the antibiotic and no-antibiotic groups, respectively, effect size 0.51, 95% CI [0.27–0.96], P=.036). There were no differences between groups with regard to occurrence of immediate postoperative fever that was not explained by a work up for infection or other etiology. The results of the matched cohort analysis similarly confirmed findings of a significantly higher rate of surgical site infection in the group that did not receive antibiotics (0 vs 7.8% of cases in antibiotic and no-antibiotic groups, respectively, P=.014) (Table 5).
Although the 221 cases involving chromopertubation were removed from the cohort before additional statistical analysis owing to the possibility of introduction of ascending genital tract pathogens into the surgical field, it was noted that chromopertubation was more frequently associated with use of perioperative antibiotics (16.3% vs 2.2% in antibiotic and no-antibiotic groups, respectively, P<.001). There were eight (3.6%) surgical site infections within the subgroup of patients who underwent chromopertubation; seven of the patients received perioperative antibiotics (five superficial surgical site infections and two organ or space surgical site infections). There was one patient who did not receive perioperative antibiotics and underwent chromopertubation, then experienced an organ or space surgical site infection.
The multivariable regression analyses of infectious outcomes controlling for age, route of surgery, high risk factors, any intraoperative complication, myoma weight, entrance into endometrial cavity are depicted in Table 6. With the antibiotic group held as reference, there was nearly a fourfold increase in odds of any surgical site infection (adjusted odds ratio [aOR] 3.77, 95% CI [1.30–10.97], P=.015), minor infection (aOR 3.74, 95% CI [1.49–9.36], P=.005) or any infectious outcome (aOR 4.36, 95% CI [1.86–10.22], P=.007) when antibiotics were not administered.
In this large retrospective cohort study, we demonstrate widespread use of perioperative antibiotics at the time of myomectomy. This is in contrast to published guidelines that do not classify myomectomy as a high-risk surgical procedure necessitating perioperative antibiotic therapy.3 We demonstrated a significant decrease in postoperative infectious outcomes, particularly surgical site infection, when perioperative antibiotics were administered. This effect was maintained despite the higher level of surgical complexity seen in the cases that received antibiotics (as reflected by significantly higher estimated blood loss, operative time, average weight of myomas removed, myoma weight, and frequency of entry into the endometrial cavity), and was corroborated on matched cohort analysis controlling for baseline differences between groups.
Perioperative antibiotic prophylaxis, advances in operative technique, and bundled interventions aimed at reducing surgical site infection rates have significantly reduced, but not eliminated infectious complications after gynecologic procedures.9,10 A case-control retrospective study of 1,531 patients undergoing hysterectomy reported a 3% surgical site infection rate despite all patients having received appropriate perioperative antibiotics; of those surgical site infection cases, 60% were classified as deep incisional or pelvic infections.11 These findings are consistent with a 2–5% infection risk quoted for patients undergoing all surgical procedures, and similar to the relatively lower rate of surgical site infection which has been reported in ambulatory surgical settings (3.1 and 4.8 per 1,000 procedures at 14 and 30 days, respectively).11–13 The incidence of surgical site infection in our antibiotic cohort, 2.8% overall and of these, 31.3% deep incisional or organ or space infection, is within range of these reports. The patients in the no antibiotics group, however, were noted to experience a higher overall surgical site infection rate of 6.8%, a third of which were deep incisional or organ or space infections. This disparity represents both a statistically and clinically significant difference in patient outcomes that may warrant a review of current management guidelines.
Notably, only 7.8% of the entire cohort did not receive antibiotics, in concordance with retrospective studies which report a high rate of antibiotic administration at time of myomectomy.4,5 The same pathogens that populate the lower genital tract and place hysterectomy at a higher risk of surgical site infection could be encountered in myomectomy, especially if chromopertubation, endometrial cavity entry or concomitant hysteroscopy is undertaken. Moreover, hematoma development is not uncommon after myomectomy and may serve as a nidus for infection, with the potential for dire consequences in a population who typically desires future fertility.
Evidence has consistently supported minimally invasive approaches to myomectomy and hysterectomy to minimize perioperative morbidity, with an accompanying 50% reduction in surgical site infection.6,14,15 After excluding the hysteroscopic cases, 70% of the myomectomies in our study were performed in minimally invasive fashion. Since the 2014 U.S. Food and Drug Administration's warnings regarding leiomyoma tissue morcellation, there has been an increase in use of alternative techniques for tissue extraction.16,17 It is unknown whether these strategies, such as mini-laparotomy, may impart additional risk of surgical site infection at time of myomectomy. Additional prospective studies will be useful to further investigate associations of surgical site infection and antibiotic usage with route of myomectomy and extraction techniques for large leiomyomas.
Strengths of this study include the large size of our population, as well as the variety of surgical routes and complexity of cases represented. It is understood that a number of factors, both modifiable and nonmodifiable, ultimately contribute to the development of an surgical site infection. We did not collect data on many such perioperative factors, including us of chlorhexidine wash, hair clipping, chlorhexidine-alcohol surgical scrub, wound closure, intraoperative temperature and technique for skin closure.18 However, given that all of the surgeries were performed at two institutions with standardized preoperative and intraoperative protocol, any variations in these practices are presumed to be nondifferential between groups. Additionally, data regarding timing of antibiotic administration during surgery was not collected. Known patient-specific risk factors for surgical site infections include diabetes, obesity, immunosuppression, smoking, cancer, cardiovascular disease, malnutrition, and prior irradiation.15,19,20 In the age group consistent with our patient population, the CDC reports that 5.1 million women in the United States are diagnosed with diabetes, 13.9% are active smokers, and 38.5% are obese with BMIs of 30 or higher.21 The generalizability of our study may be limited considering that the majority of our patient population was white and healthy with low rates of diabetes, active smoking, and an average BMI of less than 30. However, if projected onto a higher risk population that reflects the CDC's estimates on the current U.S. population, the findings of our current study have the potential to be even more pronounced. This study was performed in tertiary care centers by surgeons who were primarily subspecialists, which may additionally limit the generalizability of our findings. Owing to the retrospective observational nature of our study design, we were unable to control for the confounding effect of surgeon bias with regard to decision to administer antibiotics. There is also a possibility of data misclassifications although any errors would also be presumed to be nondifferential. Caution should be used when interpreting the striking aORs reported given the wide CIs surrounding the values, which may limit the precision of these measurements.
A high rate of antibiotic use has been demonstrated for myomectomies at our institution despite lack of supporting guidelines for this practice, and potentially reflecting the lack of prior research studies investigating the appropriateness of perioperative antibiotics in this patient population. Although the group that did not receive antibiotic prophylaxis was relatively small, a nearly fourfold increase in odds of surgical site infection was observed in the absence of antibiotic therapy. The authors suggest that these findings offer compelling evidence to support the use of perioperative antibiotics at time of abdominal, laparoscopic or robotic myomectomy. Further investigation on this topic, including prospective evaluations and cost-analyses are warranted in anticipation of the reevaluation of current guidelines on this topic. As myomectomy remains the gold standard surgical treatment for women with symptomatic myomas desiring uterine conservation, a strong evidence base is needed to guide practices that will optimize patient outcomes, improve resource utilization and assist with the global public health challenge of antimicrobial stewardship.
1. Ortega G, Rhee DS, Papandria DJ, Yang J, Ibrahim AM, Shore AD, et al. An evaluation of surgical site infections by wound classification system using the ACS-NSQIP. J Surg Res 2012;174:33–8.
2. Mittendorf R, Aronson MP, Berry RE, Williams MA, Kupelnick B, Klickstein A, et al. Avoiding serious infections associated with abdominal hysterectomy: a meta-analysis of antibiotic prophylaxis. Am J Obstet Gynecol 1993;169:1119–24.
3. Prevention of infection after gynecologic procedures. ACOG Practice Bulletin No. 195. American College of Obstetricians and Gynecologists. Obstet Gynecol 2018;131:e172–89.
4. Wright JD, Hassan K, Ananth CV, Herzog TJ, Lewin SN, Burke WM, et al. Use of guideline-based antibiotic prophylaxis in women undergoing gynecologic surgery. Obstet Gynecol 2013;122:1145–53.
5. Uppendahl L, Chiles C, Shields S, Dong F, Kraft E, Duong J, et al. Appropriate use of prophylactic antibiotic agents in gynecologic surgeries at a midwestern teaching hospital. Surg Infect 2018;19:397–402.
6. Bhave Chittawar P, Franik S, Pouwer AW, Farquhar C. Minimally invasive surgical techniques versus open myomectomy for uterine fibroids. The Cochrane Database of Systematic Reviews 2014, Issue 10. Art. No.: CD004638. DOI: 10.1002/14651858.CD004638.pub3.
7. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:205–13.
8. Berríos-Torres SI, Umscheid CA, Bratzler DW, Leas B, Stone EC, Kelz RR, et al. Centers for Disease Control and Prevention guideline for the prevention of surgical site infection, 2017. JAMA Surg 2017;152:784–91.
9. Bratzler DW, Dellinger EP, Olsen KM, Perl TM, Auwaerter PG, Bolon MK, et al. Clinical practice guidelines for antimicrobial prophylaxis in surgery. Surg Infect 2013;14:73–156.
10. Pellegrini JE, Toledo P, Soper DE, Bradford WC, Cruz DA, Levy BS, et al. Consensus bundle on prevention of surgical site infections after major gynecologic surgery. AANA J 2017;85:1–12.
11. Pop-Vicas A, Musuuza JS, Schmitz M, Al-Niaimi A, Safdar N. Incidence and risk factors for surgical site infection post-hysterectomy in a tertiary care center. Am J Infect Control 2017;45:284–7.
12. Lewis SS, Moehring RW, Chen LF, Sexton DJ, Anderson DJ. Assessing the relative burden of hospital-acquired infections in a network of community hospitals. Infect Control Hosp Epidemiol 2013;34:1229–30.
13. Owens PL, Barrett ML, Raetzman S, Maggard-Gibbons M, Steiner CA. Surgical site infections following ambulatory surgery procedures. JAMA 2014;311:709–16.
14. Young H, Knepper B, Vigil C, Miller A, Carey JC, Price CS. Sustained reduction in surgical site infection after abdominal hysterectomy. Surg Infect (Larchmt) 2013;14:460–3.
15. Gandaglia G, Ghani KR, Sood A, Meyers JR, Sammon JD, Schmid M, et al. Effect of minimally invasive surgery on the risk for surgical site infections: results from the National Surgical Quality Improvement Program (NSQIP) Database. JAMA Surg 2014;149:1039–44.
16. Harris JA, Swenson CW, Uppal S, Kamdar N, Mahnert N, As-Sanie S, et al. Practice patterns and postoperative complications before and after US Food and Drug Administration safety communication on power morcellation. Am J Obstet Gynecol 2016;214:98.e1–13.
17. Meurs E, Brito LG, Ajao MO, Goggins ER, Vitonis AF, Einarsson JI, et al. Comparison of morcellation techniques at the time of laparoscopic hysterectomy and myomectomy. J Minim Invasive Gynecol 2017;24:843–9.
18. Steiner HL, Strand EA. Surgical-site infection in gynecologic surgery: pathophysiology and prevention. Am J Obstet Gynecol 2017;217:121–8.
19. Colling KP, Glover JK, Statz CA, Geller MA, Beilman GJ. Abdominal hysterectomy: reduced risk of surgical site infection associated with robotic and laparoscopic technique. Surg Infect (Larchmt) 2015;16:498–503.
20. Krueger JK, Rohrich RJ. Clearing the smoke: the scientific rationale for tobacco abstention with plastic surgery. Plast Reconstr Surg 2001;108:1063–73.
21. National Center for Health Statistics. Health, United States, 2016: with chartbook on long-term trends in health. Hyattsville (MD): National Center for Health Statistics; 2017.