Surgical-site infections (SSIs) are the most common healthcare-associated infections in the United States; an estimated 157,500 SSIs occur annually.Reference Magill, Edwards and Bamberg 1 The Centers for Disease Control and Prevention (CDC) guidelines for the prevention of SSI recommend the use of preoperative antibiotic prophylaxis for procedures for which data support the benefit of prophylactic antibiotics, for those that involve implantation of a medical device, or in surgeries in which an SSI is potentially catastrophic.Reference Mangram, Horan, Pearson, Silver and Jarvis 2 The most recent CDC guidelines recommend against administration of prophylactic antibiotics in clean surgeries after the surgical incision is closed, even in the presence of surgical drains,Reference Berrios-Torres, Umscheid and Bratzler 3 due to lack of data showing benefit with this practice. Other national and international organizations recommend stopping prophylactic antibiotics within 24 hours of noncardiothoracic surgery.Reference Mangram, Horan, Pearson, Silver and Jarvis 2 , Reference Bratzler, Dellinger and Olsen 4 , 5 In practice, compliance with limiting the duration of prophylactic antibiotic after surgery varied by 58%–100% in a study of National Surgical Quality Improvement Program hospitals.Reference Ingraham, Cohen and Bilimoria 6 , Reference Wang, Chen, Ward and Bhattacharyya 7 A major concern with the use of postdischarge antibiotic prophylaxis is exposing patients to unnecessary antibiotics. The use of unnecessary antibiotics may result in additional costs and adverse events in addition to selection of antibiotic-resistant bacteriaReference Harbarth, Samore, Lichtenberg and Carmeli 8 and Clostridium difficile infection.Reference Poeran, Mazumdar and Rasul 9
Postdischarge antibiotic prophylaxis use is common in mastectomyReference Throckmorton, Boughey and Boostrom 10 – Reference Macadam, Clugston and Germann 13 and breast augmentation,Reference Khan 14 , Reference Perrotti, Castor, Perez and Zins 15 after which surgeons often continue antibiotics until removal of all surgical drains. The American Society of Plastic Surgeons practice guidelines for expander/implant breast reconstruction recommend that antibiotics be discontinued <24 hours after surgery unless drains are present, in which case prophylaxis duration is left to surgeon preference.Reference Alderman, Gutowski, Ahuja and Gray 16 In a survey of 460 plastic surgeons, 72% prescribed outpatient antibiotics after mastectomy with breast reconstruction.Reference Phillips, Wang and Mirrer 17 Another survey of 253 plastic surgeons who performed primarily cosmetic breast augmentation reported that 84% prescribed postoperative antibiotics, with 73% continuing them for >3 days.Reference Chopra, Gowda, McNichols, Brown, Slezak and Rasko 18
We utilized a large commercial health insurer database to analyze use and outcomes of postdischarge prophylactic antibiotics in a cohort of women undergoing mastectomy with and without immediate reconstruction. We aimed to determine the prevalence of postdischarge prophylactic antibiotics in uncomplicated procedures, to identify patient and operative factors associated with this utilization, and in a secondary analysis, to determine whether postdischarge antibiotic use was associated with decreases in SSI and noninfectious wound complications (NIWCs). The use of large generalizable health insurer claims data allowed for better understanding of practice regarding continuation of prophylactic antibiotics after hospital discharge among a wide variety of surgeons and potential stewardship opportunities to limit this practice in breast surgery. Unlike previous studies, the large database also provided the statistical power to determine the association of postdischarge antibiotics and the incident SSIs and NIWCs.
METHODS
Primary Data Source
We conducted a retrospective cohort study using all fully adjudicated, paid claims submitted for reimbursement from providers, facilities, and outpatient pharmacies linked to health plan enrollment information from 12 Anthem-affiliated plans in the HealthCore Integrated Research Database (Online Supplemental Appendix).Reference Olsen, Nickel and Fox 19 Fully insured women with enrollment in a fee-for-service plan with medical coverage of hospital and physician services and prescription drug coverage were eligible for inclusion in the cohort. Women with end-stage renal disease, prior organ transplant, or HIV infection were excluded due to the rare nature of the conditions and for privacy concerns.
Mastectomy Population
We identified inpatient mastectomy operations among insured members aged 18–64 years from January 1, 2004, to December 31, 2011, using the International Classification of Diseases, 9 th Revision, Clinical Modification (ICD-9-CM) and Current Procedural Terminology, 4 th edition (CPT-4) procedure codes from facilities and providers, as described previously.Reference Olsen, Nickel and Fox 19 Classification of the mastectomy and reconstruction has been described in detail.Reference Olsen, Nickel and Fox 19 , Reference Olsen, Nickel and Margenthaler 20 To limit the identification of antibiotics to those prescribed for prophylaxis (rather than therapeutic indications), we excluded operations for which SSI,Reference Olsen, Nickel and Fox 19 cellulitis,Reference Olsen, Nickel and Fox 19 NIWCs,Reference Nickel, Fox, Margenthaler, Wallace, Fraser and Olsen 21 port infection (ICD-9-CM diagnosis 996.62), or septicemia (ICD-9-CM diagnosis 038.0–038.9, 790.7) were coded from 30 days before mastectomy through 5 days after discharge from the mastectomy hospitalization.
Postdischarge Prophylactic Antibiotics
We defined postdischarge prophylactic antibiotics as any of the following antibiotics from paid claims in the outpatient pharmacy file from 0 to 5 days after mastectomy discharge: cephalosporins, fluoroquinolones, lincosamides, macrolides, penicillins, rifampin, sulfamethoxazole-trimethoprim, and tetracyclines.
Predictors of Prophylactic Antibiotic Use
To identify predictors of postdischarge prophylactic antibiotic use in univariate and multivariable analyses, we limited the population to (1) women with insurance coverage 365 days before surgery through 30 days after surgery, (2) the first mastectomy per patient, and (3) records without missing information on patient region of residence (Appendix figure).
Potential predictors of prophylactic antibiotic use included demographics, comorbidities (primarily using the Elixhauser classification),Reference Elixhauser, Steiner, Harris and Coffey 22 medications, cancer, operative factors, and facility factors with clinical or biologic plausibility for association with antibiotic use and/or risk for SSI, as described previously.Reference Olsen, Nickel and Margenthaler 20 , Reference Nickel, Fox, Margenthaler, Wallace, Fraser and Olsen 21 We included postoperative home health care within 10 days after discharge unless the home health visits were first coded at the time of or after an SSI or NIWC because these visits were likely due to wound care.
Complications After Mastectomy
SSIs and NIWCs first recorded from 6 days after mastectomy discharge through 90 days after mastectomy were identified using ICD-9-CM diagnosis codes from inpatient and outpatient facilities and provider claims, as reported previously.Reference Olsen, Nickel and Fox 19 , Reference Nickel, Fox, Margenthaler, Wallace, Fraser and Olsen 21 NIWCs included fat necrosis (567.82, 611.3), dehiscence (875.0, 875.1, 879.0, 879.1, 998.3, 998.32), hematoma (998.12), and tissue necrosis (998.83). We censored the observation period to capture complications before a subsequent surgical procedure, as described previously.Reference Olsen, Nickel and Fox 19 We captured C. difficile infection (ICD-9-CM diagnosis code 008.45 or prescription for oral vancomycin) through 30 days after discharge from the mastectomy hospitalization. For persons with prophylactic antibiotic use, we only included C. difficile coded on or after the first postdischarge antibiotic date.
Statistical Analyses
We determined potential predictors of prophylactic antibiotic use using χ2 tests for binary variables and a generalized estimating equations (GEE) model for multicategory variables. Any predictors with a cell size <5 were excluded from analysis. Among factors with P<.2 in univariate analysis, we assessed multicollinearity by examining the tolerance values in each model to ensure independence of explanatory variables. A GEE model was used to estimate adjusted relative risks because the prevalence of prophylactic antibiotics was high. Backward selection was used with a cutoff of P<.05. For our secondary aim, we compared the incidence of SSI and NIWCs after mastectomy according to prophylactic antibiotic use with a logistic regression model, controlling for type of breast reconstruction. All data management and analyses were performed using SAS version 9.4 software (SAS Institute, Cary, NC). We performed post hoc power calculations using PASS 14 Power Analysis and Sample Size Software (NCSS, LLC, Kaysville, UT). This study was approved by the Washington University Human Research Protection Office.
RESULTS
Starting with our previously described population of 18,696 mastectomy procedures,Reference Olsen, Nickel and Fox 19 12,501 inpatient mastectomy procedures among 12,198 patients with medical and prescription drug coverage met eligibility criteria for this study (Appendix). Immediate reconstruction was performed in 7,912 (63.3%) procedures. Prophylactic antibiotics were prescribed post discharge after 5,492 (43.9%) procedures overall. Of 5,651 unique prescriptions within 5 days post discharge, cephalosporins were the most commonly prescribed antibiotic class (75.1%), followed by fluoroquinolones (11.1%), lincosamides (4.8%), and penicillins (3.5%). Utilization of postdischarge prophylactic antibiotics was more common after mastectomy with immediate reconstruction than after mastectomy only (4,439 [56.1%] vs 1,053 [22.9%]; P<.001).
Overall, 793 procedures were coded for SSI (6.3%) and 802 (6.4%) were coded for an NIWC within 90 days after surgery. A total of 13 (0.1%) procedures included evidence for C. difficile infection within 30 days after hospital discharge. In mastectomy-only operations, the SSI rates were similar regardless of postdischarge prophylactic antibiotic use, but the rate of NIWCs was higher in women given postdischarge prophylactic antibiotics than in women who did not receive postdischarge antibiotics (Figure 1). Postdischarge prophylactic antibiotic use was not associated with SSI or NIWCs following mastectomy with immediate reconstruction (Figure 1). After adjusting for type of procedure (ie, mastectomy only, implant reconstruction, autologous flap reconstruction), receipt of postdischarge prophylactic antibiotic was not associated with SSI (odds ratio [OR], 0.87; 95% confidence interval [CI], 0.75–1.01) nor NIWCs (OR, 1.14; 95% CI, 0.99–1.33) compared to no prescription for postdischarge antibiotics. Based on the number of reconstruction procedures and assuming a 25% reduction in SSI associated with postdischarge antibiotic use, we achieved 81% power to detect a difference in the complication rates in the reconstruction population but much lower power (18%) in the mastectomy-only population due to the lower percentage receiving postdischarge antibiotics and the lower baseline complication rates.
FIGURE 1 Incidence of surgical site infection and noninfectious wound complications by postdischarge prophylactic antibiotic use and presence of immediate breast reconstruction (N=12,501). NOTE. Abx, antibiotic; NIWC, noninfectious wound complication; SSI, surgical-site infection.
The population for our analysis of risk factors associated with postdischarge antibiotic utilization was reduced to 9,188 after excluding those with <1 year of prior health insurance coverage, those with second mastectomy procedures, and 23 procedures with missing US region (Appendix figure). In univariate analysis, women who received postdischarge prophylactic antibiotics were younger, in the higher-income quartiles, and more likely to have preexisting depression, tobacco use disorder, skin disease, breast carcinoma in situ or to be undergoing prophylactic mastectomy than women who did not receive postdischarge antibiotics (Table 1). Women who received postdischarge antibiotics were also more likely to have undergone sentinel node dissection, implant, autologous flap, and autologous flap plus implant reconstruction, and less likely to have had a modified radical mastectomy than women who did not receive antibiotics post discharge. Several facility factors were associated with increased likelihood to receive postdischarge prophylactic antibiotics, including larger bed size, medical school affiliation, residency program, and facility location in the US Northeast and Midwest regions (Table 1).
TABLE 1 Univariate Predictors of Postdischarge Prophylactic Antibiotics After 9,188 Mastectomy ProceduresFootnote a
NOTE. Ref, reference group.
a The following factors had P≥0.2 and were excluded from the table: alcohol abuse, blood loss anemia, coagulopathy, collagen vascular/ rheumatologic disease, congestive heart failure, deficiency anemia, disease-modifying anti-rheumatic drugs, oral steroids, peripheral vascular disease, pneumonia or urinary tract infection within 30 days prior to surgery, psychoses, renal failure, prior Staphylococcus aureus infection, inflammatory breast disease, history of breast cancer, and history of radiotherapy.
b Based on median for zip code.
c Missing facility information due to no facility claim for procedure, no match to a facility in the American Hospital Association Annual Survey of Hospitals, or a match to multiple facilities.
d For those with missing facility information, we used patient home zip code when possible to determine the US region of the facility.
e Restricted to home health care before a surgical site infection or noninfectious wound complication.
The independent predictors of prophylactic antibiotic use post discharge with the largest effect sizes were operative factors: immediate breast implant (relative risk [RR], 2.41) and autologous flap reconstruction with an implant (RR, 2.17) or without an implant (RR, 2.11). Patient factors (hypertension, tobacco use disorder, and postoperative home health) and facility characteristics (surgery at a facility in the Northeast or Midwest and surgery at an academic hospital) were associated with slightly increased risk of postdischarge prophylactic antibiotic use after controlling for the type of procedure (Table 2).
TABLE 2 Multivariable Predictors of Postdischarge Prophylactic Antibiotics After Mastectomy Performed During an Inpatient Hospitalization
DISCUSSION
We determined variation and predictors of postdischarge prophylactic antibiotic use in a large population of women undergoing mastectomy using geographically diverse claims data. Despite recommendations in many surgical guidelines to discontinue administration of prophylactic antibiotics either immediately after surgery or within 24 hours after a procedure, antibiotics are frequently overused postoperatively, particularly after breast operations. We found that utilization of postdischarge prophylactic antibiotics ranged from 23% for mastectomy without immediate reconstruction to 56% for mastectomy with immediate reconstruction. These results are in line with survey data suggesting that plastic surgeons are more likely to continue patients on prophylactic antibiotics post discharge than general breast surgeons.Reference Phillips, Wang and Mirrer 17 , Reference Brahmbhatt, Huebner and Scow 23 Given their common utilization of antibiotics, surgeons are positioned to be key partners in the success of hospital antibiotic stewardship programs.Reference Sartelli, Duane and Catena 24
Factors associated with prolonged perioperative prophylaxis are critical to development of interventions to assure appropriate antibiotic use. By far, the most important predictors of postdischarge prophylactic antibiotics after mastectomy were autologous flap and implant reconstruction, which were associated with a 2.1–2.4-fold increased risk of prolonged antibiotic use. Immediate reconstruction is preferred over delayed reconstruction by many plastic surgeons due to better cosmesis and perceived psychosocial benefits for the patient.Reference Serletti, Fosnot, Nelson, Disa and Bucky 25 Surgery performed in a teaching facility and in facilities located in the US Northeast and Midwest regions were also associated with a slightly increased risk of postdischarge prophylactic antibiotic use. This regional variation in the use of prolonged antibiotics is important to investigate further because it may have broad implications for regional differences in antibiotic stewardship programs as well as rates of antibiotic resistance in the community. In contrast, patient characteristics were associated with only slightly increased risk of prolonged antibiotic utilization, suggesting that stewardship interventions to reduce postdischarge antibiotic utilization should target primarily plastic surgeons performing immediate reconstruction.
In addition, 3 prior observational breast surgery studies reported significantly lower SSI rates in patients given postdischarge prophylactic antibiotics compared to either the recommended practice of a single dose of antibiotic before incisionReference Clayton, Bazakas, Lee, Hultman and Halvorson 26 or prophylaxis limited to 24 hours after surgery.Reference Edwards, Stukenborg, Brenin and Schroen 27 , Reference Avashia, Mohan, Berhane and Oeltjen 28 The study by Edwards et al included procedures performed by only 2 surgeons, 1 of whom always prescribed postoperative antibiotics until drains were removed while the other surgeon never used postoperative antibiotics. Therefore, confounding bias may be present in this study due to unmeasured differences in operative technique or other practices between these 2 surgeons.Reference Edwards, Stukenborg, Brenin and Schroen 27 In the studies by Clayton et alReference Clayton, Bazakas, Lee, Hultman and Halvorson 26 and Avashia et al,Reference Avashia, Mohan, Berhane and Oeltjen 28 retrospective comparisons of SSI rates during time periods with different prophylactic antibiotic practices were performed. It is possible that other infection control and operative practices varied between time periods, resulting in residual confounding. In the study by Avashia et al,Reference Avashia, Mohan, Berhane and Oeltjen 28 a change in prophylactic antibiotic practice was prompted by a 31.6% infection rate in 19 women during the period when only perioperative antibiotics were used, raising the possibility that the decreased SSI rate after reverting back to prolonged antibiotic administrative was due to regression to the mean. The 5 observational studiesReference Throckmorton, Boughey and Boostrom 10 , Reference Liu, Dubbins, Louie, Said, Neligan and Mathes 29 – Reference Drury, Lanier and Khavanin 32 and 1 randomized trialReference Phillips, Fourman and Bishawi 33 in the literature that did not find significantly different SSI rates associated with postdischarge antibiotic use compared to 24 hours or less of antibiotic administration were small and thus underpowered to detect a significant difference if it existed.Reference Olsen, Nickel and Fox 34 The lack of high-quality data on the risk and benefits of postdischarge antibiotic prophylaxis likely perpetuates the continued use of inappropriate prolonged antibiotic prophylaxis beyond the recommended 24 hours of duration.
In contrast to the 3 aforementioned studies that reported lower SSI rates associated with postdischarge prophylactic antibiotic use, we found that the rates of SSI and NIWCs did not differ according to use of postdischarge prophylactic antibiotics after controlling for procedure type. Importantly, we had sufficient power (>80%) in the reconstruction population to detect a difference between postdischarge prophylactic antibiotic use and the incidence of complications.Reference Throckmorton, Boughey and Boostrom 10 , Reference Clayton, Bazakas, Lee, Hultman and Halvorson 26 – Reference Phillips, Fourman and Bishawi 33 Cephalosporins were the antibiotic class most commonly prescribed for postdischarge prophylaxis in our cohort, followed by fluoroquinolones. Both of these antibiotic classes are associated with selection for colonization by and infection with antibiotic-resistant bacteria.Reference Paterson 35 Fluoroquinolones are also increasingly reported to be associated with serious adverse events.Reference Wise, Peloquin, Choi, Lane and Zhang 36 – Reference Etminan, Brophy and Samii 38
The coding algorithms we used based on ICD-9-CM and CPT-4 codes have relatively good sensitivity to identify SSI after breast proceduresReference Olsen and Fraser 39 and C. difficile infection.Reference Dubberke, Butler and Yokoe 40 Despite this, our estimates of SSI, NIWCs, and C. difficile infection rates are almost certainly underestimates of the true incidence because the claims algorithm is unlikely to detect most minor infections, especially during the global 90-day major-surgery reimbursement period. It is possible that not all postdischarge antibiotic use in the absence of diagnosis codes for infection or NIWCs represented continuation of antibiotic prophylaxis, although this is unlikely because our window to identify postdischarge prophylaxis was very short (within 5 days after discharge). Outpatient antibiotics were identified by paid pharmacy claims; thus, we could not detect antibiotics given as samples or in the absence of a prescription drug claim, and we could not confirm that antibiotics were taken as directed. Although facility characteristics were missing for approximately one-third of procedures, we think differential misclassification is unlikely, such that certain types of facilities would be disproportionally more or less likely to have a paid claim in the database or match to the AHA database. We were, however, able to impute the facility region by using patient zip code, which was important because regional utilization of postdischarge antibiotics varied.
The strengths of our study include the large population of women undergoing mastectomy, resulting in sufficient power to detect differences in complication rates in the reconstruction population. Our population was enriched in women undergoing breast reconstruction because of the use of commercial claims data because younger women are more likely to undergo immediate reconstruction.Reference Jagsi, Jiang and Momoh 41 Additional strengths include the ability to detect regional differences in antibiotic use because of the broad geographic coverage in the insurer claims data, and analysis of variation in use of prophylactic antibiotics by operative, patient, and facility characteristics.
In conclusion, postdischarge prophylactic antibiotic use in the absence of infectious or noninfectious wound complications is common after mastectomy, and it is driven primarily by immediate reconstruction rather than patient-level characteristics. We found no benefit to continuation of prophylactic antibiotics post discharge, with similar SSI and NIWC complication rates in women who received postdischarge antibiotics versus women who did not receive continued prophylactic antibiotics. Stewardship efforts to limit the duration of prophylactic antibiotics after surgery and to discourage continuation of antibiotics after hospital discharge are essential to avoiding further increases in antimicrobial resistance.
ACKNOWLEDGMENTS
We thank Cherie Hill for database and computer management support. The findings and conclusions in this document are those of the authors, who are responsible for its content, and they do not necessarily represent the official view of the NIH, the CDC, or the AHRQ.
Financial support: Funding for this project was provided by the National Institutes of Health (NIH; grant no. 5R01CA149614 to M.A.O.). Additional support was provided by the Centers for Disease Control and Prevention (CDC) Epicenters Program (grant no. U54CK000162 to V.J.F.), a grant from the National Center for Advancing Translational Sciences (NCATS) of the NIH (grant no. UL1 TR000448), and a grant from the Agency for Healthcare Research and Quality (AHRQ; grant no R24 HS19455).
Potential conflicts of interest: M.A.O. reports consultant work with Merck, Pfizer, and Sanofi Pasteur and grant funding through Pfizer, and Sanofi Pasteur for work outside the submitted manuscript. A.E.W. is an employee of HealthCore, a wholly owned subsidiary of Anthem, Inc., a health insurance company; she has received Anthem stock options and has participated in an Anthem employee stock purchase plan. V.J.F. reports that her spouse is the Senior Vice President and Chief Medical Officer for Express Scripts; she has received grants from the Foundation for Barnes-Jewish Hospital, the CDC, and the Doris Duke Foundation. D.K.W. reports consultant work with Centene Corp., Worrell, Inc., Novaerus, and Carefusion; he is a subinvestigator for a Cepheid-sponsored study for work outside the submitted manuscript. No other authors report conflicts of interest relevant to this article.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https:/doi.org/10.1017/ice.2017.128
