Use of Antimicrobial Prophylaxis for Major Surgery:  Baseline Results From the National Surgical Infection Prevention Project FREE

Surgical site infections (SSIs) are a major contributor to patient injury, mortality, and health care costs. Of the nearly 30 million operations in the United States each year, more than 2% are complicated by an SSI.^1- 5 Mortality rates are 2 to 3 times higher in patients in whom an SSI develops compared with uninfected patients, and hospital readmission rates are significantly increased.^6- 8 Surgical site infections increase length of stay by an average of 7 days and charges by approximately $3000.^6,9 When an SSI is diagnosed after hospital discharge, outpatient and emergency department visits, use of radiology services, hospital readmissions, and use of home health care services increase significantly, as do total health care costs.¹⁰ The attributable costs of an SSI may be more than $30 000 for patients undergoing cardiac or orthopedic operations.^11- 14

The effectiveness of antimicrobials administered shortly before skin incision for prevention of SSIs was established in the 1960s and has been repeatedly demonstrated since.^15- 21 However, despite evidence of effectiveness and the publication of guidelines for antimicrobial prophylaxis,^22- 28 use is often suboptimal. Previous studies have demonstrated inappropriate timing of antimicrobial administration, inappropriate selection of the antimicrobial agent, and excess duration of prophylaxis.^29- 34 In a 1996 study of patients undergoing abdominal aortic aneurysm repair, hip replacement, or large-bowel resection in 44 hospitals in New York State, Silver and colleagues³⁰ demonstrated that 14% did not receive antimicrobial prophylaxis, and only 63% of those who received antimicrobials had them administered in the 2 hours before incision.

In 2002, the Centers for Medicare & Medicaid Services, in collaboration with the Centers for Disease Control and Prevention, implemented the National Surgical Infection Prevention Project. The project promotes prophylactic practices that have been shown to reduce the risk of SSI, and thus reduce morbidity and mortality in the Medicare population. This project builds on experience that the Centers for Disease Control and Prevention has gained from implementation of the National Nosocomial Infections Surveillance System^35- 38 and on previous efforts by Centers for Medicare & Medicaid Services to improve health care quality through its Medicare quality improvement organizations.^39- 41 This report describes baseline results of an assessment of use of antimicrobial prophylaxis for a national sample of Medicare patients undergoing 5 types of major surgery during 2001.

After a literature review, a panel of experts in surgical infection prevention, hospital infection control, and epidemiology developed 3 performance measures for national surveillance and quality improvement.⁴² These measures include (1) the proportion of patients in whom parenteral antimicrobial prophylaxis is initiated within 1 hour before the incision (2 hours for vancomycin hydrochloride); (2) the proportion of patients who are given a prophylactic antimicrobial regimen consistent with published guidelines^22- 26,28; and (3) the proportion of patients whose antimicrobial prophylaxis is discontinued within 24 hours after the end of surgery. Timing and selection of antimicrobial prophylaxis were selected as measures because of the association with reduced SSI incidence. Duration was selected because excessive use of antimicrobials promotes bacterial resistance. Operations were selected for inclusion in the project based on frequency in the Medicare population, rate of postoperative SSI, and consensus regarding appropriate antimicrobial prophylaxis. Informed consent and institutional review board approval were not required because access to these data for quality assurance is given to the Medicare program by statute.

The study population included Medicare fee-for-service beneficiaries from all states or territories except the Virgin Islands who underwent selected operations from January 1 through November 30, 2001. A case was defined by a Medicare part A claim with a principal or secondary procedure code from the International Classification of Diseases, Ninth Revision, Clinical Modification,⁴³ for coronary artery bypass grafting (36.10-36.17, 36.19, and 36.2), open-chest cardiac surgery (35.0-35.51, 35.53-35.95, 35.98, and 35.99), colon surgery (45.00, 45.03, 45.41, 45.49, 45.50, 45.7-45.90, 45.92-45.95, 46.03, 46.04, 46.1-46.14, 46.52, 46.75, 46.76, 46.91, 46.92, 46.94, 48.5, and 48.6-48.69), hip or knee arthroplasty (81.51, 81.52, and 81.54), abdominal or vaginal hysterectomy (68.3, 68.4, and 68.5-68.7), or vascular surgery (38.14, 38.16, 38.18, 38.34, 38.36, 38.37, 38.44, 38.48, 38.49, 38.51, 38.52, 38.64, 39.25, 39.26, and 39.29).⁴² A systematic random sample of 788 cases was selected for each state or territory to give approximately 750 usable records after allowing for exclusions and unavailable charts. In states with fewer than 750 eligible medical charts based on the selected operations, all available discharges were reviewed.

Hospitals sent photocopies of medical records to 1 of 2 clinical data abstraction centers. Abstractors used computerized tools with explicit predefined criteria to record data elements, including patient demographics, diagnosis and procedure codes, antimicrobial use before surgery, documented antimicrobial allergies, presence of any preoperative infection, operating room arrival time, anesthesia start time, surgical incision time or surgery start time, and surgery end time. Detailed information on route, administration time of all doses, and time of discontinuation were collected for each course of antimicrobial prophylaxis administered during the hospitalization. Data on postoperative complications, including infections during the hospitalization, were collected if documented in the medical chart.

A subset of 559 cases was randomly selected for an independent review by a second abstractor. The κ statistics for the 3 performance measures included 0.76 for antimicrobials received within 1 hour before incision, 0.91 for antimicrobial selection consistent with published guidelines, and 0.86 for antimicrobial prophylaxis discontinued within 24 hours after the end of surgery, indicating high interabstractor agreement.

Cases were excluded from analysis if results of the medical chart review did not confirm that 1 of the operations of interest was performed or if there were missing dates or times that prevented evaluation of antimicrobial administration timing. Because the project focuses on prophylactic antimicrobial use, cases with documented preoperative infection, preadmission antimicrobial administration (except for oral antimicrobial prophylaxis for colon surgery), or more than 24 hours of preoperative antimicrobial administration were excluded.

The proportion of patients receiving antimicrobial prophylaxis within 60 minutes (120 minutes for vancomycin) before incision was determined by comparing the start time of the last parenteral antimicrobial dose before the surgical incision with the documented incision time. Cases without an explicitly documented incision time were excluded from this measure. We also excluded cases with insufficient timing data or if a patient undergoing colon surgery received oral antimicrobial prophylaxis only.

The proportion of patients receiving antimicrobial drugs consistent with published guidelines was calculated by evaluating whether the patient received any antimicrobial or combination of antimicrobials consistent with any of the guidelines.⁴² Antimicrobials were considered prophylactic if given before surgery, intraoperatively, or within 24 hours of the end of surgery. Cases were excluded from this measure if no antimicrobials were administered or if no antimicrobials were considered prophylactic. Because there were no guidelines for antimicrobial selection in patients with a β-lactam allergy undergoing colon surgery or hysterectomy, these patients were excluded from this measure.

The proportion of patients whose antimicrobial prophylaxis was discontinued within 24 hours after the end of surgery was calculated by comparing the start time of the last dose of prophylactic antimicrobial with the surgery end time. Cases were excluded from this measure if any infection was documented intraoperatively or within 48 hours after the end of surgery. In addition, cases were excluded if no antimicrobials were administered, if no antimicrobials were considered prophylactic, if there were insufficient data to determine antimicrobial timing, or if the patient underwent more than 1 operation of interest during the hospitalization.

Because the sample of cases represented a fixed maximum number of patients per state, we applied normalized weights to adjust all performance measures. We calculated exact binomial 95% confidence intervals (CIs) for unweighted and weighted performance measure rates. All analyses were completed using SAS statistical software (SAS version 8.2; SAS Institute Inc, Cary, NC).

We abstracted data from a total of 39 086 medical records and applied general exclusions sequentially. Excluded cases included 205 (0.5%) with no surgery of interest performed, 1817 (4.7%) with documentation of preoperative infection, 2 (<0.01%) with all antimicrobial dates and times missing, 1461 (3.7%) with preadmission antimicrobial administration, 1432 (3.7%) with antimicrobial administration for more than 24 hours preoperatively, and 36 cases (0.09%) for other reasons. After exclusions, 34 133 (87.3%) cases from 2965 hospitals (median, 5 cases per hospital; range, 1-386 cases) remained available for analysis.

Patient demographic characteristics are summarized in Table 1. The mean age was 73.3 years (SD, 9.4 years); 89.3% were white; and 58.6% were female. Hip or knee arthroplasty made up the largest group (44%) of operations evaluated. The median length of stay for all patients was 5 days (interquartile range, 4-8 days), and the in-hospital mortality rate was 2.7% (95% CI, 2.5%-2.9%).

Only 232 patients (0.7%) did not receive a prophylactic antimicrobial agent. The proportion of patients who received an antimicrobial dose within 60 minutes (120 minutes for vancomycin) before the incision is summarized in Table 2. An explicit incision time was documented for only 11 220 patients (32.9%). Of these, 55.7% (95% CI, 54.8%-56.6%) received an antimicrobial dose in the recommended time frame. We randomly selected and reabstracted 1728 cases for which an explicit incision time was missing. Using proxy times (eg, surgery start time) to approximate the incision time, we estimated that 54.3% of these patients received their antimicrobial dose within 1 hour before incision. Patients undergoing cardiac and orthopedic surgery were more likely to receive an antimicrobial dose within 1 hour before incision. This may reflect the much more common use of preprinted care plans or order forms (50.0% of patients undergoing hip or knee arthroplasty and 36.6% of patients undergoing cardiac surgery), which often included antimicrobial protocols. Preprinted care plans or order forms were found in only 4.0% of colorectal surgery cases, 4.5% of hysterectomy cases, and 5.3% of vascular surgery cases. The distribution of antimicrobial start times related to incision is demonstrated in Figure 1. We noted that 9.6% of patients received their first antimicrobial dose more than 4 hours after incision. For those patients whose operation was still under way more than 4 hours after incision, 12.2% received an additional intraoperative antimicrobial dose.

The proportion of patients who received an antimicrobial drug consistent with published guidelines is summarized in Table 3. Most patients (92.6%; 95% CI, 92.3%-92.8%) received a guideline-recommended antimicrobial regimen. The most commonly prescribed agents for patients who received any antimicrobial or combination of antimicrobials consistent with surgical antimicrobial prophylaxis guidelines are summarized in Table 4. First- or second-generation cephalosporins were the most commonly used agents in those cases that met the criteria. In cases of cardiac and vascular surgery and total joint arthroplasty, the antimicrobial most commonly considered inappropriate was vancomycin because of no documentation of a β-lactam allergy. Combinations of ampicillin sodium and sulbactam sodium and of piperacillin sodium and tazobactam sodium, antimicrobials not included in any guidelines for surgical prophylaxis, were the most commonly used drugs in colon surgery cases that failed the performance measure. The proportion of patients who received prophylaxis that was limited to a guideline-recommended regimen (ie, no additional antimicrobials were given beyond those recommended) was 78.7%.

The proportion of cases in which prophylaxis was discontinued within 24 hours after the end of surgery is shown in Table 5 and Figure 2. Only 40.7% (95% CI, 40.2%-41.2%) of patients had their antimicrobial prophylaxis discontinued within 24 hours. Patients undergoing cardiac surgery (34.4%; 95% CI, 33.4%-35.5%) and hip or knee arthroplasty (36.7%; 95% CI, 35.9-37.4) were least likely to have antimicrobial prophylaxis discontinued, although the median duration of antimicrobial prophylaxis was longest (57.0 hours) for patients undergoing colon surgery. At 48 hours after the end of surgery, 26.7% of patients were still receiving prophylaxis, and 9.3% continued to receive prophylaxis for more than 96 hours (Figure 2).

Postoperative infections were documented during hospitalization in 2875 patients (8.4%). Pneumonia or other lung infection was documented in 1235 patients (3.6%); urinary tract infections, in 974 (2.9%); bloodstream infections, in 541 (1.6%); and SSIs in 357 (1.0%). Rates of SSIs by procedure type were 1.3% for cardiac surgery, 2.9% for colon surgery, 0.2% for hip or knee arthroplasty, 0.4% for hysterectomy, and 1.7% for vascular surgery. The most commonly documented noninfectious complications included hemorrhage in 1411 patients (4.1%), heart failure or pulmonary edema in 1400 (4.1%), respiratory failure in 1282 (3.8%), and cardiac arrest in 556 (1.6%).

Antimicrobial prophylaxis to prevent SSI is one of the most widely accepted practices in surgery. Optimal prophylaxis requires application in appropriate types of operations, selection of safe and effective antimicrobials, initial administration and redosing to maintain effective serum and tissue levels throughout the operation, and discontinuation when the patient is no longer receiving a benefit.^5,17,23,25 The results of this population-based study of selected operations performed on Medicare beneficiaries suggest that essentially all patients (99.3%) receive antimicrobial prophylaxis, but that there is still considerable room for improvement in its timing and discontinuation.

Overall, 55.7% of patients received prophylactic antimicrobials during the 60 minutes (120 minutes for vancomycin) before incision. Prior studies have demonstrated that timing is critical to the effectiveness of prophylaxis, and current guidelines recommend dosing within 1 hour before incision. In 1957, Miles et al¹⁵ demonstrated that measures that influenced the development of experimental infections were effective only during the 3 to 4 hours after bacterial contamination of a wound. In 1961, Burke¹⁶ reported that a variety of antimicrobials could prevent the development of experimental infections, but only if given within about 3 hours after wound contamination. In randomized clinical trials reported in 1964 and 1969, antimicrobials given before, during, and shortly after abdominal surgery were effective in preventing SSI.^17,44 Stone et al¹⁸ showed that the lowest rates of SSI in abdominal operations were associated with prophylaxis started within 1 hour before the incision, whereas Classen et al²¹ reported similar findings for cardiac operations in 1992. It is of interest that 9.6% of the patients in our assessment received their first antimicrobial dose more than 4 hours after incision when little if any benefit would be expected based on these previously published studies.^16,18,21

Most (92.6%) of the patients in this assessment received a prophylactic antimicrobial regimen consistent with current guidelines. However, only 78.7% received regimens that were limited to recommended agents, suggesting that a substantial amount of antimicrobials are used unnecessarily.

Many different antimicrobial regimens are effective for preventing SSI. The optimal agent is effective against the organisms that are most likely to be encountered during the type of operation that is to be performed; is safe, inexpensive, and bactericidal; and has a long half-life.²⁵ Several professional organizations have published guidelines for the selection and use of prophylactic antimicrobials.^22- 28 These guidelines generally favor the use of older, relatively narrow-spectrum agents such as the first- and second-generation cephalosporins. There is no evidence that newer antimicrobials are more effective than older options. The prophylactic use of newer, broad-spectrum drugs that are frontline therapeutic agents is generally discouraged because of concern that widespread use will promote the emergence and spread of bacterial strains that are resistant to them. Our data suggest that newer antimicrobials are frequently used when older agents would be effective.

The results of this study raise additional concerns regarding antimicrobial resistance. Our data suggest that vancomycin continues to be used excessively for surgical prophylaxis. The primary indication for use of prophylactic vancomycin is a β-lactam allergy. In almost 50% of the cases in this study where vancomycin was used, there was no documented β-lactam allergy. In addition, 59.3% of patients received prophylaxis for more than 24 hours after the end of surgery. The optimal duration of prophylaxis has been controversial, especially for cardiovascular and orthopedic surgery, where many surgeons prefer to continue prophylaxis until all drains and tubes have been removed. However, the bulk of published evidence, including studies of orthopedic and cardiac surgery, suggests that short-duration prophylaxis—as little as 1 dose in many studies—is equally effective as longer-duration administration in preventing infection.^{5,22- 28,45- 62} There is also evidence that prolonged antimicrobial administration can be harmful to patients by promoting antimicrobial-resistant bacteria and increasing the incidence of antibiotic-associated complications.^63- 71

Our study has several limitations. We only included Medicare patients undergoing 5 major surgical procedures, and our findings may not apply to younger patients or those undergoing other operations. Our findings were limited to information documented in the medical record, and important case findings may have been omitted. This was especially apparent in the lack of an explicitly documented incision time in almost two thirds of the charts. We also had no information on rates of methicillin-resistant Staphylococcus aureus infection in individual hospitals and no information on prior methicillin-resistant S aureus colonization of patients undergoing surgery. This prevented us from determining if there might be legitimate reasons for use of vancomycin in some patients without a documented β-lactam allergy. Vancomycin might be the agent of choice for certain operations when a hospital has high rates of infection with methicillin-resistant S aureus or methicillin-resistant coagulase-negative staphylococci or when a patient has known colonization with these organisms. Unfortunately, an institutional threshold rate that should trigger the use of vancomycin has not been established.²⁵ Routine use of vancomycin for prophylaxis is not justified for any operation type because of the risk of promoting vancomycin-resistant bacteria in the patient and hospital environment. Finally, data on postoperative infections were limited to those documented in the chart. No information about nosocomial infection surveillance or postdischarge complications was available for this study, thereby limiting our ability to compare SSI incidence with existing national data (eg, the National Nosocomial Infections Surveillance System of the Centers for Disease Control and Prevention).³⁸

The findings of this study continue to document the challenges of disseminating evidence-based knowledge systematically and expeditiously into clinical practice.^72- 73 The scientific basis for antimicrobial prophylaxis was first published more than 40 years ago.^15- 17 When considering the broad range of evidence-based recommendations for clinical practice, appropriate surgical antimicrobial prophylaxis would appear to be a relatively straightforward practice to implement. However, many patients continue to undergo operations without having received antimicrobials during the critical period before incision, and prolonged duration of antimicrobial prophylaxis remains common. Although the demand for high quality and accountability in health care is increasing, greater effort is needed to ensure that proven treatments are routinely implemented in practice. In the recent Institute of Medicine report, Priority Areas for National Action: Transforming Health Care Quality,⁷⁴ prevention and surveillance of nosocomial infections was recommended as a priority area for health care quality improvement. The authors noted that “prevention of nosocomial infections through the implementation of evidence-based guidelines and a surveillance program demonstrates how effective systems interventions can lead to more widespread improvements in quality of care.” One option to improve surgical antimicrobial prophylaxis and prescribing may be broader development and use of computer-assisted order entry and decision support.^75- 76

The findings from this baseline evaluation represent the first step in a number of interventions designed to improve the delivery of surgical care in the United States. Medicare quality improvement organizations across the country have implemented local quality improvement initiatives with hospitals to improve delivery of antimicrobial prophylaxis for surgical patients.⁴² In addition, the Joint Commission on Accreditation of Healthcare Organizations has adopted the 3 National Surgical Infection Prevention Project performance measures as an ORYX core measure initiative (Joint Commission on Accreditation of Healthcare Organizations, Oakbrook Terrace, Ill) that will be used in the future to focus on-site survey evaluation activities.⁷⁷ The performance measures related to timing, selection, and duration of antimicrobial prophylaxis have been slated for voluntary submission and public reporting of hospital performance beginning in the summer of 2005.⁷⁸

Substantial opportunities remain to improve the use of prophylactic antimicrobials for patients undergoing major surgery. Achieving high rates of performance for appropriate antimicrobial prophylaxis to prevent SSIs will require the development of systems in which the knowledge from years of research and recommendations from clinical practice guidelines are routinely incorporated into practice.

Correspondence: Dale W. Bratzler, DO, MPH, Oklahoma Foundation for Medical Quality, Inc, 14000 Quail Springs Pkwy, Suite 400, Oklahoma City, OK 73134-2627 (dbratzler@okqio.sdps.org).

Accepted for Publication: July 27, 2004.

Disclaimer: The analyses on which this publication is based were performed under contract 500-99-P619, entitled “Utilization and Quality Control Peer Review Organization for the State of Oklahoma,” sponsored by the Centers for Medicare & Medicaid Services, Department of Health and Human Services. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, and the mention of trade names, commercial products, or organizations does not imply endorsement by the US government. The authors assume full responsibility for the accuracy and completeness of the ideas presented. This article is a direct result of the Health Care Quality Improvement Program initiated by CMS, which has encouraged identification of quality improvement projects derived from analyses of patterns of care, and therefore required no special funding on the part of this contractor. Ideas and contributions to the authors concerning experience in engaging with issues presented are welcomed.

Acknowledgment: We thank the following professionals who have served on the expert panel for the National Surgical Infection Prevention Project: Jason H. Calhoun, MD, and Arlen D. Hanssen, MD (American Academy of Orthopaedic Surgery, Rosemont, Ill); Justine Medina, RN, MS (American Association of Critical Care Nurses, Aliso Viejo, Calif); Sandra Tunajek, CRNA, MSN (American Association of Nurse Anesthetists, Park Ridge, Ill); Vanessa Dalton, MD, MPH (American College of Obstetricians and Gynecologists, Washington, DC); Philip S. Barie, MD, and Christopher Daly, MD (American College of Surgeons, Chicago, Ill); Robert A. Bonomo, MD (American Geriatrics Society, New York, NY); Carl W. Armstrong, MD (American Hospital Association, Chicago); James E. Cottrell, MD, Lee A Fleisher, MD, and Jonathon R. Gavrin, MD (American Society of Anesthesiologists, Park Ridge); Neil H. Hyman, MD (American Society of Colon and Rectal Surgeons, Arlington Heights, Ill); Keith M. Olsen, PharmD (American Society of Health-System Pharmacists, Bethesda, Md); Leslie Shultz, PhD, CPHQ (Ascension Health, St Louis, Mo); Carol Peterson, RN, BSN, MAOM, CNOR (Association of periOperative Registered Nurses, Denver, Colo); Audrey Adams, RN, MPH, CIC (Association for Professionals in Infection Control and Epidemiology, Washington); Julie Gerberding, MD, MPH, Chesley Richards, MD, MPH, and Lynn Steele, MS, CIC (Centers for Disease Control and Prevention, Atlanta, Ga); Peter M. Houck, MD, and Judy Goldfarb, RN, MA (CMS, Seattle, Wash); James T. Lee, MD, PhD (Healthcare Infection Control Practices Advisory Committee, Atlanta); E. Patchen Dellinger, MD, and Peter Gross, MD (Infectious Diseases Society of America, Alexandria, Va); Sharon Sprenger, RHIA, CPHQ, MPA (Joint Commission on Accreditation of Healthcare Organizations, Oakbrook Terrace, Ill); Gianna Zuccotti, MD, MPH (The Medical Letter, New Rochelle, NY); Dale W. Bratzler, DO, MPH, Karina Carr, RN, and Lisa Red, MSHA (Oklahoma Foundation for Medical Quality, Inc, Oklahoma City); Meera Kelly, MD, George Parkins, MD, Gina Pugliese, RN, MS, Judene Bartley, RN, MPH, and Colleen M. Vetere, RN, MPH (Premier, Inc, Oak Brook, Ill); Rosa Johnson, ARNP, MN (Qualis Health, Seattle); William R. Jarvis, MD, and Edward S. Wong (Society for Healthcare Epidemiology of America, Alexandria); Fred H. Edwards, MD, and Mary Eiken (Society of Thoracic Surgeons, Chicago); Donald E. Fry, MD (Surgical Infection Society, Cincinnati, Ohio); and John A. Hitt, MD (VHA, Inc, Denver).

Financial Disclosure: Dr Fry is a member of the speakers’ bureaus of and has received research grants from Pfizer and Merck. Dr Dellinger has received grants for clinical research from, served on an advisory board for, and/or lectured for honoraria from SmithKline Beecham, Glaxo, Bayer, Eli Lilly, Merck, Wyeth-Ayerst, Pharmacia, Bristol-Myers Squibb, AstraZeneca, Pfizer, Aventis, Hoffmann–La Roche, Arrow, NABI, Ortho-McNeil, Parke-Davis, Abbott, ICOS, Immunex, Chiron, Searle, Cubist, Versicor, InterMune, Peninsula, BRAHMS, and Centocor.