Abstract
OBJECTIVE
To evaluate associations between prophylactic preoperative antibiotic choice and surgical site infection rates after hysterectomy.
METHODS
A retrospective cohort study was performed of patients in the Michigan Surgical Quality Collaborative undergoing hysterectomy from July 2012 to February 2015. The primary outcome was a composite outcome of any surgical site infection (superficial surgical site infections or combined deep–organ space surgical site infections). Preoperative antibiotics were categorized based on the recommendations set forth by the American College of Obstetricians and Gynecologists and the Surgical Care Improvement Project. Patients receiving a recommended antibiotic regimen were categorized into those receiving beta-lactam antibiotics and those receiving alternatives to beta-lactam antibiotics. Patients receiving non-recommended antibiotics were categorized into those receiving overtreatment (excluded from further analysis) and those receiving non-standard antibiotics. Multivariable logistic regression models were developed to estimate the independent effect of antibiotic choice. Propensity score matching analysis was performed to validate the results.
RESULTS
The study included 21,358 hysterectomies. The overall rate of any surgical site infection’ was 2.06% (N=441). Unadjusted rates of ‘any surgical site infection’ were 1.8%, 3.1% and 3.7% for beta-lactam, beta-lactam alternatives and non-standard groups, respectively. After adjusting for patient and operative factors within clusters of hospitals, compared to the beta-lactam antibiotics (reference group), the risk of ‘any surgical site infection’ was higher for the group receiving beta-lactam alternatives (OR 1.7, CI 1.27–2.07) or the non-standard antibiotics (OR 2.0, CI 1.31–3.1).
CONCLUSION
Compared to women receiving beta lactam antibiotic regimens, there is a higher risk of surgical site infection after hysterectomy among those receiving a recommended beta lactam alternative or non-standard regimen.
Surgical site infections are associated with increased hospital length of stay and increased episode cost after surgery.1,2 In addition, surgical site infections are the most common reason for readmission after a wide variety of operations.3 The rate of overall surgical site infections (superficial, deep, and organ space) in hysterectomy has been reported to range between 1% and 4%.4,5 Hysterectomy is among the most common major operations in the United States (600,000 performed annually) and may result in 6,000 to 24,000 surgical site infections each year. Consequently, beginning October 1, 2015, inpatient post-hysterectomy surgical site infections were included in the Centers for Medicare and Medicaid Services calculations for Hospital-Acquired Condition Reduction Program metrics.6
Preoperative prophylactic antibiotic administration has been shown to consistently reduce the rate of postoperative surgical site infections.7 The American College of Obstetricians and Gynecologists (the College) has issued guidelines for choosing appropriate preoperative prophylactic antibiotics8; in addition, the Joint Commission’s Surgical Care Improvement Project has issued a list of procedure-specific prophylactic antibiotics.9 Studies have shown that compliance with these guidelines varies across institutions and procedures,10 and regimens not in compliance have involved both undertreated and overtreated cohorts.
The objective of the current study is to quantify the effects of preoperative antibiotic choice on surgical site infection rates after hysterectomy using data from a statewide surgical collaborative.
MATERIALS AND METHODS
A retrospective cohort study was performed of patients in the Michigan Surgical Quality Collaborative (herein also referred to as the collaborative) undergoing hysterectomy from July 2012 to February 2015. The collaborative is funded by the Blue Cross Blue Shield of Michigan/Blue Care Network, and it includes patients from all insurance payers (public and private). At each participating hospital, a trained, dedicated nurse abstractor collects patient characteristics, intraoperative processes of care (including the details of preoperative antibiotics administered), and 30-day postoperative outcomes from general and vascular surgery and hysterectomy cases. To ensure complete capture of the data, nurse abstractors make phone calls to the patients to determine whether they were admitted to a hospital other than the one in which the index surgery was performed. To reduce sampling error, a standardized data collection methodology is used whereby data abstraction is performed on only the first 25 cases of an 8-day cycle (alternating on different days of the week for each cycle). The standardized data collection methodology is routinely validated through scheduled site visits, conference calls, and internal audits. 11,12
Patients were included in the study if they were older than 18 years of age and were undergoing abdominal, vaginal, laparoscopic, or robotic hysterectomy. Patients with gynecologic malignancy and those undergoing hysterectomy for benign indications were included in the study. Patients with no recorded antibiotic information and those with missing surgical site infection information were excluded from the analysis. Michigan Surgical Quality Collaborative datasets provided to the researchers contain no patient, hospital or healthcare provider identifiers. Therefore, this study met the criteria for “exempt” status by the University of Michigan Institutional Review Board–Medical (HUM00073978).
The following information was available for analysis: age at the time of surgery; body mass index (BMI), calculated as weight/height (kg/m2); covariates associated with performance status, including American Society of Anesthesiologists (ASA) classification score (defined as a dichotomous variable an ASA class <3 or ≥3)13; and preoperative medical history, including diabetes mellitus (defined as requiring oral hypoglycemic agents and/or insulin), hypertension (defined as documentation in preoperative evaluation or of receiving antihypertension medications), and smoking status (defined as having smoked cigarettes, cigars, or pipe, chewed tobacco, or used marijuana within the past year). Preoperative transfusion was defined as having receiving a minimum of one unit of whole blood or packed red blood cells during the 72 hours before surgery. Patients with a final diagnosis coded as 179 to 184 based on the primary International Classification of Diseases, Ninth Revision (ICD-9) were defined as having the diagnosis of gynecologic cancer. All other ICD-9 diagnoses were defined as benign final pathology.
Approach to hysterectomy was categorized as open (all abdominal hysterectomy cases and all cases converted from laparoscopic or robotic cases) or minimally invasive, which encompassed laparoscopic (including robotic-assisted cases) and vaginal (including laparoscopic-assisted cases). Surgical complexity was calculated by adding the relative value units for each surgical procedure recorded for the patient. Operative times were reported in hours, from the start of the surgery (incision) to the closing of the skin incision.
Surgical site infections within 30 days of surgery were defined by the Centers for Disease Control and Prevention criteria. A superficial surgical site infection involved only skin and the subcutaneous tissue of the incision. In this study, deep and organ space surgical site infections were both considered “deep surgical site infections” because the fascia and muscle layers of the vaginal cuff are contiguous with the organ space. The primary outcome of the study was a composite outcome of any surgical site infection. The term any surgical site infection indicates when there was either a superficial or deep surgical site infection.
Preoperative antibiotics were categorized based on the criteria set forth by the the American College of Obstetricians and Gynecologists 8 and the Surgical Care Improvement Project.9 Patients receiving an antibiotic regimen recommended by ACOG or the Surgical Care Improvement Project were further categorized into those receiving beta-lactam antibiotics (e.g. cephalosporin, ampicillin-sulbactam, ertapenem) and those receiving alternatives to beta-lactam antibiotics (e.g. combination of clindamycin with gentamicin or quinolone). Patients receiving antibiotic regimens not recommended by the ACOG or the Surgical Care Improvement Project were categorized as those receiving overtreatment (e.g. recommended antibiotic with additional antibiotic) and those receiving non-standard antibiotics (e.g. Clindamycin alone). Patients who received overtreatment were excluded from the analysis because documented antibiotic resistance could account for such a decision. Figure 1 illustrates the development of the antibiotic categories.
Fig. 1.
Breakdown of the antibiotic categories based on the American College of Obstetricians and Gynecologists (the College) usage guidelines and the Surgical Care Improvement Project (SCIP) usage guidelines. Antibiotic regimen details available in Appendix 2, available online at http://links.lww.com/xxx. *Patients receiving additional antibiotics to those recommended by the College and SCIP guidelines were categorized as overtreatment.
For all included patients, descriptive and comparative statistics of demographics, comorbidities, operative details, and postoperative surgical site infections were analyzed. For bivariate analyses, chi-square analysis or Fisher’s exact test was used. For continuous variables, parametric one-way analysis of variance or nonparametric Wilcoxon Mann-Whitney tests were used to assess significance in the bivariate relationship. To ascertain the independent effect of antibiotic categories included in the analysis, we constructed multivariate logistic regression models. Variables were excluded from model selection if they were not significant at a level of 0.1 in the bivariate analysis or if they were not related to the outcome in a clinically plausible manner.
For all logistic regression models, to account for violations in model assumptions due to non-independence of observations within clusters of data (hospital level), we used Huber-Eicker-White robust standard errors. These robust standard errors and the hospital-level clustering allowed the model to better reflect the collected data characteristics.13–15 We used STATA 14.0 SE for Macintosh (StataCorp LP, College Station, TX) for all analyses. Results of the logistic regression models were confirmed using propensity score matching (Appendix 1, available online at http://links.lww.com/xxx).
RESULTS
A total of 22,992 patients undergoing hysterectomy were available in the collaborative database. Excluded from the analysis were cases with no recorded antibiotic information (n=418; 1.8%) and with missing surgical site infection information (n=29; 0.1%). Patients who received overtreatment were excluded from the analysis (n=1,187; 5.1%). There were 21,358 (93%) who were included in the analysis (Fig 1).
Most of these patients received beta-lactam antibiotics (N=17,827; 79.1%), followed by the beta-lactam alternatives (N=2,878; 12.8%). The non-standard regimens were administered in 2.8% (N=653) of cases (Fig. 1). The majority of patients in the non-standard group received single agent antibiotics (Clindamycin alone − 67%; Gentamicin only − 8%) Details of the 15 different regimens included in this group are provided in Appendix 2, available online at http://links.lww.com/xxx.
The overall rate of any surgical site infection was 2.06% (N=441). Patients with any surgical site infection were older, had higher BMI, were more likely to have diabetes, were more likely to report tobacco use, received a preoperative transfusion, and had gynecologic cancer as a surgical indication. In addition, patients with any surgical site infection had higher use of open abdominal approach, higher median blood loss, higher complexity of surgery (measured by mean relative value units), and longer operative times (Table 1). Baseline comparison between the three groups of antibiotic categories is provided in Table 2. Patients receiving a beta-lactam antibiotics had lower incidence of tobacco use, ASA class 3, history of hypertension, and history of diabetes. The three groups did not differ in the operative time, blood loss, surgical complexity, and proportion of patients with malignancy. The beta-lactam antibiotics group had a higher proportion of patients undergoing open surgery than the other two groups. Unadjusted surgical site infection rates were 1.8% for beta-lactam antibiotics, 3.1% for beta-lactam alternatives, and 3.75% for non-standard antibiotics. Details of the unadjusted rates of any surgical site infection, superficial surgical site infections, and deep surgical site infections are provided in Table 2.
Table 1.
Predictors of Surgical Site Infection (unadjusted)
| Surgical Site Infection (any)
|
||||
|---|---|---|---|---|
| Variable | Overall N=22,358 |
Absent N=22,917 |
Present N=441 |
P |
|
| ||||
| Demographics and Comorbidities | ||||
|
| ||||
| Age, y, median (SD) | 48.1 (11.7) | 48.1 (11.7) | 47.9 (11.8) | .8 |
|
| ||||
| BMI, N (%) | ||||
| ≥30 kg/m2 obese | 10,150 (47.5) | 9,879 (97.3) | 271 (2.7) | <.001 |
| <30 kg/m2 nonobese | 11,208 (52.5) | 11,038 (98.5) | 170 (1.5) | |
|
| ||||
| Diabetes, N (%) | ||||
| Present | 1,984 (8.8) | 1,911 (96.3) | 73 (3.7) | <.001 |
| Absent | 20,561 (91.2) | 20,169 (98.1) | 392 (1.9) | |
|
| ||||
| Smoker, N (%) | ||||
| Yes | 4,991 (23.4) | 4,864 (97.5) | 127 (2.5) | .006 |
| No | 16,367 (76.6) | 16,053 (98.1) | 314 (1.9) | |
|
| ||||
| ASA class, N (%) | ||||
| ≤2 | 16,812 (78.7) | 16,514 (98.2) | 298 (1.8) | <.001 |
| ≥3 | 4,546 (21.2) | 4,403 (96.8) | 143 (3.2) | |
|
| ||||
| History of hypertension, N (%) | ||||
| Present | 6,358 (30) | 6,209 (97.2) | 176 (2.8) | <.001 |
| Absent | 14,973 (70) | 14,708 (98.2) | 265 (1.8) | |
|
| ||||
| Preoperative transfusion, N (%) | ||||
| Yes | 146 (0.7) | 139 (95.2) | 7 (4.8) | .02 |
| No | 21,212 (99.3) | 20,778 (98) | 434 (2) | |
|
| ||||
| Final pathology, N (%) | ||||
| Cancer | 1,997 (9.3) | 1,911 (95.7) | 86 (4.3) | <.001 |
| Benign | 19,261 (90.7) | 19,006 (98.2) | 355 (1.8) | |
|
| ||||
| Surgical Factors | ||||
|
| ||||
| Surgical approach, N (%) | ||||
| Open | 5,797 (27.1) | 5,569 (96.1) | 228 (3.9) | <.001 |
| Minimally invasive* | 15,561 (72.9) | 15,348 (98.6) | 213 (1.4) | |
|
| ||||
| Estimated blood loss, mL, median (IQR) | 100 (50–200) | 100 (50–200) | 200 (100–350) | <.001 |
|
| ||||
| Surgical complexity, total RVU, mean (SD) | 26.7 (14.1) | 26.5 (13.9) | 31.8 (21.7) | <.001 |
|
| ||||
| Operative time, h, median (SD) | 2.2 (1.3) | 2.1 (1) | 2.5 (1.3) | <.001 |
|
| ||||
| Antibiotic Type | ||||
|
| ||||
| Beta-lactam antibiotics, N (%) | 17,827 (83.5) | 17,498 (98.2) | 329 (1.8) | <.001 |
|
|
||||
| Beta-lactam alternatives, N (%) | 2,878 (13.5) | 2,790 (96.9) | 88 (3.1) | |
|
|
||||
| Non-standard | 653 (3) | 629 (96.3) | 24 (3.7) | |
ASA, American Society of Anesthesiologists; BMI, body mass index; PCN, penicillin; RVU, relative value units.
Laparoscopic, vaginal, and robotic hysterectomy.
Table 2.
Baseline Comparison of Characteristics Between the Antibiotic Groups
| Variable | Beta-lactam antibiotics N=17,827 |
Beta-lactam alternatives N=2,878 |
Non-standard N=653 |
P |
|---|---|---|---|---|
| Demographics and Comorbidities
| ||||
| Age, y, median (SD) | 48 (11.4) | 48.5 (12.3) | 48 (12.2) | .09 |
| BMI, kg/m2, mean (SD) | 30.8 (8) | 31.6 (8) | 31.7 (9) | .001 |
| Diabetes present, N (%) | 1,461 (8.2) | 305 (10.6) | 68 (10.4) | .001 |
| Tobacco user, N (%) | 4,088 (22.9) | 730 (25.4) | 173 (26.5) | .003 |
| ASA class ≥3, N (%) | 3,632 (20.4) | 744 (25.9) | 170 (26) | <.001 |
| History of hypertension, N (%) | 5,245 (29.4) | 931 (32.3) | 209 (32) | .03 |
| Preoperative transfusion, N (%) | 122 (0.7) | 23 (0.8) | 1 (0.2) | .1 |
| Gynecologic cancer, N (%) | 1,624 (9.1) | 310 (10.8) | 63 (9.6) | <.001 |
|
| ||||
| Perioperative Factors
| ||||
| Surgical approach, open, N (%) | 4,880 (27.4) | 757 (26.3) | 160 (24.5) | <.001 |
| Estimated blood loss, mL, median (IQR) | 100 (50–200) | 100 (50–200) | 100 (50–199) | .6 |
| Surgical complexity, total RVU, mean (SD) | 25.9 (12.8) | 26.8 (13.4) | 25.4 (13.4) | <.001 |
| Operative time, h, mean (SD) | 2.2 (1) | 2.2 (1) | 2 (1) | .2 |
|
| ||||
| Surgical Site Infection (unadjusted)
| ||||
| Any, N (%) | 329 (1.8) | 88 (3.1) | 24 (3.7) | <.001 |
| Superficial, N (%) | 165 (0.9) | 41 (1.4) | 15 (2.3) | <.001 |
| Deep–organ, N (%) | 167 (0.9) | 47 (1.6) | 10 (1.5) | .003 |
ASA, American Society of Anesthesiologists; BMI, body mass index; IQR, interquartile range; PCN, penicillin; RVU, relative value units.
Multivariate logistic regression models were constructed for any surgical site infection, superficial surgical site infections, and deep surgical site infections. Table 3 summarizes the independent effect of factors included in the regression models. Compared to the beta-lactam antibiotics (reference group), patients receiving the beta-lactam alternatives had increased risk of any surgical site infection (odds ratio [OR] 1.62, 95% confidence interval [CI] 1.27–2.07, P<.001), superficial surgical site infections (OR 1.5, 95% CI 1.04–2.09, P=.03), and deep–organ space surgical site infections (OR 1.7, 95% CI 1.27–2.4, P<.001). Similarly, compared to the beta-lactam antibiotics (reference group), patients receiving any non-standard regimen had at least twice the risk of any surgical site infection (OR 2.0 95% CI 1.31–3.1, P<.001), superficial surgical site infections (OR 2.5, 95% CI 1.46–4.34, P<.001) but did not differ significantly in the rate of deep–organ space (Table 3). The adjusted rate of any surgical site infection with respect to the antibiotic categories is shown in Figure 2. Results of the logistic regression were validated using propensity score matching (Appendix 1, http://links.lww.com/xxx).
Table 3.
Logistic Regression Model: Independent Predictors of Surgical Site Infection)
| Variable adjusted for in logistic regression model | Any SSI | Superficial SSI | Deep SSI | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||||||
| Unadjusted OR | Adjusted OR | 95% CI | P | Unadjusted OR | Adjusted OR | 95% CI | P | Unadjusted OR | Adjusted OR | 95% CI | P | |||||
|
| ||||||||||||||||
| Antibiotic category | ||||||||||||||||
| Beta-lactam antibiotics | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| Beta-lactam alternatives | 1.7 | 1.62 | 1.27–2.07 | <.001 | 1.6 | 1.5 | 1.04–2.09 | .03 | 1.8 | 1.7 | 1.2–2.4 | <.001 | ||||
| Non-Standard | 2.1 | 2.02 | 1.31–3.1 | <.001 | 2.5 | 2.5 | 1.46–4.34 | .001 | 1.7 | 1.6 | 0.8–3.1 | .1 | ||||
|
| ||||||||||||||||
| Surgical time (per hour) | 1.3 | 1.23 | 1.14–1.33 | <.001 | 1.3 | 1.14 | 0.98–1.3 | .08 | 1.3 | 1.26 | 1.14 –1.39 | <.001 | ||||
|
| ||||||||||||||||
| BMI | ||||||||||||||||
| <30 kg/m2 nonobese | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| ≥30 kg/m2 obese | 1.8 | 1.5 | 1.2–1.9 | <.001 | 2.3 | 1.8 | 1.2–2.7 | <.001 | 1.4 | 1.2 | 0.8–1.6 | .2 | ||||
|
| ||||||||||||||||
| Smoking status | ||||||||||||||||
| Nonsmoker | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| Smoker | 1.46 | 1.18–1.8 | <.001 | 1.5 | 1.7 | 1.2–2.2 | <.001 | 1.2 | 1.27 | 0.9–1.7 | .14 | |||||
|
| ||||||||||||||||
| ASA category | ||||||||||||||||
| <3 | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| ≥3 | 1.8 | 1.12 | 0.9–1.4 | .3 | 2.2 | 1.2 | 0.8–1.7 | .5 | 1.4 | 1.1 | 0.8–1.5 | .5 | ||||
|
| ||||||||||||||||
| Surgical complexity (per RVU) | 1.02 | 1.01 | 0.9–1.01 | .13 | 1.02 | 1.01 | 0.9–1.02 | .1 | 1.01 | 1.002 | 0.9–1.01 | .9 | ||||
|
| ||||||||||||||||
| Diabetes | ||||||||||||||||
| Absent | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| Present | 1.8 | 1.36 | 1.05–1.76 | .02 | 2.2 | 1.5 | 1.04–2.2 | .03 | 1.4 | 1.3 | 0.8–1.8 | .24 | ||||
|
| ||||||||||||||||
| Final pathology | ||||||||||||||||
| Benign | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| Cancer | 2.4 | 1.7 | 1.3–2.2 | <.001 | 3 | 1.8 | 1.3–2.6 | .01 | 1.8 | 1.5 | 0.9–2.5 | .08 | ||||
|
| ||||||||||||||||
| Surgical route | ||||||||||||||||
| MIS | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | Ref | ||||
| Open | 3 | 2.6 | 2.1–3.1 | <.001 | 4.2 | 3.5 | 2.6–5 | <.001 | 2 | 1.9 | 1.37–2.6 | <.001 | ||||
ASA, American Society of Anesthesiologists; BMI, body mass index; CI, confidence interval; MIS, minimally invasive surgery (vaginal, laparoscopic, and robotic hysterectomies); OR, odds ratio; PCN, penicillin; Ref, referent category; RVU, relative value units.
Fig. 2.
Adjusted rates of overall surgical site infection by antibiotic category. Rates adjusted for patient factors (age, body mass index, American Society of Anesthesiologists category, history of diabetes, gynecologic malignancy, and tobacco use) and operative factors (surgical time, blood loss, and surgical complexity). Red dashed lines indicate 95% confidence interval bounds for referent category.
The overall rate of non-standard antibiotics uses in the collaborative dropped from 5.2% to 2.5% over the study time period. (Fig. 3)
Fig. 3.
Proportion of patients over time enrolled in the Michigan Surgical Quality Collaborative receiving nonstandard antibiotic regimens per the American College of Obstetricians and Gynecologists usage guidelines and the Surgical Care Improvement Project usage guidelines.
DISCUSSION
In this retrospective analysis of patients undergoing hysterectomy in the Michigan Surgical Quality Collaborative, we found that the choice of antibiotic regimen given before hysterectomy independently predicts the rate of any surgical site infection. Beta-lactam antibiotics (cephalosporins, ampicillin-sulbactam, ertapenem) are associated with the lowest rates of surgical site infections. Recommended beta-lactam alternatives (e.g. Clindamycin plus Gentamicin or Quinolone or Aztreonam) and patients receiving non-standard regimens (e.g. Gentamicin only, Clindamycin only) have a significantly higher risk of surgical site infections. One possible explanation is that beta-lactam antibiotics are highly effective against skin flora (Streptococcus species, Staphylococcus aureus, and coagulase-negative staphylococci), which are the predominant organisms that cause surgical site infections.16–18 Regimens that do not contain a beta-lactam antibiotic are inferior in controlling these organisms. 7
Given this increased risk, patient-reported allergy to penicillin (PCN) should be thoroughly investigated to ascertain its validity and severity. Previous studies have shown that because of the fear of PCN anaphylaxis, clinicians frequently accept a diagnosis of PCN allergy without obtaining a detailed history of the reaction. 19 In our study, approximately 12% of the patients received a beta-lactam alternative antibiotic regimen, a prevalence consistent with the self-reported PCN allergy described in the literature. 20 It is important to remember that cephalosporin cross-reactivity shown in skin testing is present in only 10% of patients with a true PCN allergy. 19,21 Patients with negative results on PCN skin testing and those without a history of anaphylactic reaction to PCN can safely receive cephalosporin. 19,22,23 Routine use of PCN skin testing could potentially increase the use of cephalosporins and therefore reduce the use of alternative antibiotics in perioperative settings. 24,25
The current analysis quantifies the association of administering antibiotics not recommended by the College or by Surgical Care Improvement Project guidelines before hysterectomy. Wright et al10 reported that 2.3% of patients undergoing gynecologic surgery received antibiotics not recommended by the guidelines. However, the authors did not report the effect of non-adherence to guidelines on surgical site infection rates. In our study, the majority of patients who received a non-standard regimen received a single-agent antibiotic (clindamycin, gentamicin, or metronidazole). Previous studies have shown these single agents are inferior to cephalosporins. 26 Studies have also shown that adherence to Surgical Care Improvement Project’s surgical site infection reduction bundle into surgical safety checklists can significantly improve antibiotic infusion timing and antibiotic selection. 27 In our study, for each quarter starting in July 2012, the percentage of patients in the collaborative who receive non-standard antibiotics has consistently decreased (Fig. 3). Although precise reasons of this improvement are likely multifactorial, the participation of hospitals in a functional collaborative encouraging evidence-based practices seems to improve the quality of surgical care across the hospitals. 28
This study has several strengths. The Michigan Surgical Quality Collaborative is a statewide collaborative that uses standardized data collection methods and dedicated nurse abstractors who are regularly audited for interrater reliability. Although the collaborative is limited to a single state, it includes a mix of academic and community hospitals, making the data more generalizable. In addition, our logistic regression modeling accounted for the clustering effect from physician/facility preferences. Studies have shown that the quality of data from collaboratives such as the Michigan Surgical Quality Collaborative and the National Surgical Quality Improvement Project is similar to that of chart review and much better than that of administrative claims-based databases. 29
Limitations of our study include reported heterogeneity in surgical site infection reporting in the literature; however, collaborative abstractors are trained to reduce variations in reporting. Moreover, the Centers for Disease Control and Prevention criteria for surgical site infection diagnosis may underestimate the true incidence of surgical site infections by excluding cases of cellulitis by as much as threefold. 30 Even though the nurse abstractors follow up with patients by phone within the 30-day period to avoid missing capturing complications if patients seek care in another hospital, potential for underreporting surgical site infections remains. Lastly, data on the appropriate timing and dosage of antibiotics were not available, and variations in these could have affected the conclusions of this study.
In summary, efforts to decrease surgical site infections could focus on adherence to recommended preoperative antibiotic guidelines and thorough evaluation of patient reported penicillin allergies in order to increase the number of patients receiving beta lactam antibiotics.
Supplementary Material
Acknowledgments
The authors thank Marie Chesny and Sarah Block for their editorial assistance. Investigator support for C.W.S. was provided by the National Institute of Child Health and Human Development WRHR Career Development Award # K12 HD065257. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Financial Disclosure
The authors did not report any potential conflicts of interest.
References
- 1.Schweizer ML, Cullen JJ, Perencevich EN, Vaughan Sarrazin MS. Costs associated with surgical site infections in Veterans Affairs hospitals. JAMA Surg. 2014;149:6. doi: 10.1001/jamasurg.2013.4663. Available at: http://dx.doi.org/10.1001/jamasurg.2013.4663. Retrieved October 13, 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Young MH, Washer L, Malani PN. Surgical site infections in older adults: epidemiology and management strategies. Drugs Aging. 2008;25:399–414. doi: 10.2165/00002512-200825050-00004. [DOI] [PubMed] [Google Scholar]
- 3.Merkow RP, Ju MH, Chung JW, Hall BL, Cohen ME, Williams MV, et al. Underlying reasons associated with hospital readmission following surgery in the United States. JAMA. 2015;313:483–95. doi: 10.1001/jama.2014.18614. [DOI] [PubMed] [Google Scholar]
- 4.Lake AG, McPencow AM, Dick-Biascoechea MA, Martin DK, Erekson EA. Surgical site infection after hysterectomy. Am J Obstet Gynecol. 2013;209:490.e1–9. doi: 10.1016/j.ajog.2013.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Olsen MA, Higham-Kessler J, Yokoe DS, Butler AM, Vostok J, Stevenson KB, et al. Developing a risk stratification model for surgical site infection after abdominal hysterectomy. Infect Control Hosp Epidemiol. 2009;30:1077–83. doi: 10.1086/606166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Centers for Medicaid & Medicare Services. [Accessed October 13, 2015];CMS to improve quality of care during hospital inpatient stays. Available at: http://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2014-Fact-sheets-items/2014-08-04-2.html.
- 7.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. doi: 10.1089/sur.2013.9999. [DOI] [PubMed] [Google Scholar]
- 8.ACOG Committee on Practice Bulletins--Gynecology. ACOG practice bulletin no 104: antibiotic prophylaxis for gynecologic procedures. Obstet Gynecol. 2009;113:1180–9. doi: 10.1097/AOG.0b013e3181a6d011. [DOI] [PubMed] [Google Scholar]
- 9.The Joint Commission. [October 13, 2015];Specifications manual for national hospital inpatient quality measures. Available at: http://www.jointcommission.org/specifications_manual_for_national_hospital_inpatient_quality_measures.aspx.
- 10.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. doi: 10.1097/AOG.0b013e3182a8a36a. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Corona LE, Swenson CW, Sheetz KH, Shelby G, Berger MB, Pearlman MD, et al. Use of other treatments before hysterectomy for benign conditions in a statewide hospital collaborative. Am J Obstet Gynecol. 2015;212:304e1–7. doi: 10.1016/j.ajog.2014.11.031. [DOI] [PubMed] [Google Scholar]
- 12.Hendren S, Fritze D, Banerjee M, Kubus J, Cleary RK, Englesbe MJ, et al. Antibiotic choice is independently associated with risk of surgical site infection after colectomy: a population-based cohort study. Ann Surg. 2013;257:469–75. doi: 10.1097/SLA.0b013e31826c4009. [DOI] [PubMed] [Google Scholar]
- 13.King G, Roberts ME. How robust standard errors expose methodological problems they do not fix, and what to do about it. Political Analysis. 2015;23:159–79. [Google Scholar]
- 14.White H. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica. 1980;48:817–38. [Google Scholar]
- 15.Rabe-Hesketh S, Everitt B, editors. A Handbook of Statistical Analyses Using Stata. 4. London: Chapman & Hall/CRC; 2007. Maximum likelihood estimation. [Google Scholar]
- 16.Dumville JC, McFarlane E, Edwards P, Lipp A, Holmes A. Preoperative skin antiseptics for preventing surgical wound infections after clean surgery. Cochrane Database Syst Rev. 2013;3:CD003949. doi: 10.1002/14651858.CD003949.pub3. [DOI] [PubMed] [Google Scholar]
- 17.Sievert DM, Ricks P, Edwards JR, Schneider A, Patel J, Srinivasan A, et al. Antimicrobial-resistant pathogens associated with healthcare-associated infections: summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2009–2010. Infect Control Hosp Epidemiol. 2013;34:1–14. doi: 10.1086/668770. [DOI] [PubMed] [Google Scholar]
- 18.Hidron AI, Edwards JR, Patel J, Horan TC, Sievert DM, Pollock DA, et al. NHSN annual update: antimicrobial-resistant pathogens associated with healthcare-associated infections: annual summary of data reported to the National Healthcare Safety Network at the Centers for Disease Control and Prevention, 2006–2007. Infect Control Hosp Epidemiol. 2008;29:996–1011. doi: 10.1086/591861. [DOI] [PubMed] [Google Scholar]
- 19.Salkind AR, Cuddy PG, Foxworth JW. Is this patient allergic to penicillin? An evidence-based analysis of the likelihood of penicillin allergy. JAMA. 2001;285:2498–505. doi: 10.1001/jama.285.19.2498. [DOI] [PubMed] [Google Scholar]
- 20.Raja AS, Lindsell CJ, Bernstein JA, Codispoti CD, Moellman JJ. The use of penicillin skin testing to assess the prevalence of penicillin allergy in an emergency department setting. Ann Emerg Med. 2009;54:72–7. doi: 10.1016/j.annemergmed.2008.12.034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Surtees SJ, Stockton MG, Gietzen TW. Allergy to penicillin: fable or fact? BMJ. 1991;302:1051–2. doi: 10.1136/bmj.302.6784.1051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Herbert ME, Brewster GS, Lanctot-Herbert M. Medical myth: ten percent of patients who are allergic to penicillin will have serious reactions if exposed to cephalosporins. West J Med. 2000;172:341. doi: 10.1136/ewjm.172.5.341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Haslam S, Yen D, Dvirnik N, Engen D. Cefazolin use in patients who report a non-IgE mediated penicillin allergy: a retrospective look at adverse reactions in arthroplasty. Iowa Orthop J. 2012;32:100–3. [PMC free article] [PubMed] [Google Scholar]
- 24.Cook DJ, Barbara DW, Singh KE, Dearani JA. Penicillin skin testing in cardiac surgery. J Thorac Cardiovasc Surg. 2014;147:1931–5. doi: 10.1016/j.jtcvs.2014.01.019. [DOI] [PubMed] [Google Scholar]
- 25.Frigas E, Park MA, Narr BJ, Volcheck GW, Danielson DR, Markus PJ, et al. Preoperative evaluation of patients with history of allergy to penicillin: comparison of 2 models of practice. Mayo Clin Proc. 2008;83:651–62. doi: 10.4065/83.6.651. [DOI] [PubMed] [Google Scholar]
- 26.Brummer THI, Heikkinen A-M, Jalkanen J, Fraser J, Mäkinen J, Tomás E, et al. Antibiotic prophylaxis for hysterectomy, a prospective cohort study: cefuroxime, metronidazole, or both? BJOG. 2013;120:1269–76. doi: 10.1111/1471-0528.12178. [DOI] [PubMed] [Google Scholar]
- 27.Tillman M, Wehbe-Janek H, Hodges B, Smythe WR, Papaconstantinou HT. Surgical care improvement project and surgical site infections: can integration in the surgical safety checklist improve quality performance and clinical outcomes? J Surg Res. 2013;184:150–156. doi: 10.1016/j.jss.2013.03.048. [DOI] [PubMed] [Google Scholar]
- 28.Campbell DA, Jr, Englesbe MJ, Kubus JJ, Phillips LRS, Shanley CJ, Velanovich V, et al. Accelerating the pace of surgical quality improvement: the power of hospital collaboration. Arch Surg. 2010;145:985–91. doi: 10.1001/archsurg.2010.220. [DOI] [PubMed] [Google Scholar]
- 29.Lawson EH, Louie R, Zingmond DS, Brook RH, Hall BL, Han L, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg. 2012;256:973–81. doi: 10.1097/SLA.0b013e31826b4c4f. [DOI] [PubMed] [Google Scholar]
- 30.Degnim AC, Throckmorton AD, Boostrom SY, Boughey JC, Holifield A, Baddour LM, et al. Surgical site infection after breast surgery: impact of 2010 CDC reporting guidelines. Ann Surg Oncol. 2012;19:4099–103. doi: 10.1245/s10434-012-2448-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.