William R. Jarvis
Historical Background
Until the end of the 19th century, infection was the greatest risk associated with any surgical procedure. Although the practice of surgery had spread rapidly following the introduction of ether at Massachusetts General Hospital in 1846, even relatively minor procedures could be complicated by severe systemic infection and death, which for major procedures was the expected outcome. Increasing knowledge regarding the relationship between bacteria and infection, advanced at the end of the century by such legendary figures as Pasteur, Lister, and Koch, led to a series of discoveries and the development of techniques that ultimately paved the way for modern surgery. Pasteur studied the relationship between bacteria and putrefaction. Lister recognized the role of bacteria in surgical wound infections and in 1867 introduced the practice of spraying antiseptics into wounds to combat bacteria. He gave scant attention, however, to the role of hands in introducing bacteria into surgical wounds. In an early (1878) application of his own postulates concerning microbial pathogens, Koch produced experimental wound infections by the injection of bacteria. Subsequently, in 1881, he verified the superiority of heat over antiseptics for killing bacteria and preventing access of bacteria to wounds [1].
The role of the surgeon's hands in introducing bacteria into wounds was slow to be recognized despite the work of Semmelweis in 1847. In 1882, when Ernst Bergmann arrived at the Ziegelstrasse Clinic in Berlin and was asked what was new in surgery, he replied, “Today we wash our hands before operation” [1]. Although rubber gloves were first developed for the use of Halsted's scrub nurse in 1889 to protect her hands from harsh antiseptics, widespread use of rubber gloves in surgical procedures did not become established until well into the 20th century.
In the 20th century, the standardization of aseptic practices in the operating room greatly improved the safety of clean operative procedures, but operations involving anatomic structures with a dense endogenous flora that cannot be eliminated before operation, such as the colon and rectum, continued to carry a very high risk of infection. A major collaborative study organized by the National Research Council (NRC) in 1964 documented the rate of surgical site infection (SSI) following 15,613 operations carried out over 27 months from 1959 to 1962 in 16 operating rooms of five university hospitals [2]. The study was designed to investigate whether the reduction of airborne bacteria in the operating room, accomplished with ultraviolet light irradiation, could achieve a reduction in SSIs. The study found that ultraviolet light produced a significant reduction in airborne bacteria but had no effect on SSIs except in the class of “refined clean” wounds (i.e., those with the lowest probability for contamination by endogenous bacteria). The infection rates in all other classes of wounds—other (clean), clean-contaminated, contaminated, and dirty—were unaffected by ultraviolet irradiation.
The NRC study was one of the earliest and certainly one of the most convincing to document the importance of endogenous bacteria as the primary etiologic agent of SSIs. This report also introduced a system for classifying wounds according to the risk of endogenous contamination (and thus of postoperative wound infection), which provided a basis for comparing SSI statistics. Although more sensitive and specific wound classification systems employing additional risk factors for wound infection have been developed since the NRC study [3,4], all systems continue to incorporate elements of this original scheme. Its NRC report contained results from the largest and most carefully conducted study in its day to examine a host of other factors related to the patient and the environment that influenced the risk of postoperative wound infections.
P.584
Multivariate analysis of this large body of data provided convincing evidence of changes in the risk of developing postoperative infections influenced by the patient's age, obesity, steroid administration, malnutrition, presence of remote infection, use of drains, duration of operation, and duration of preoperative hospitalization. It is interesting to note that although diabetic patients had a higher infection rate than did nondiabetics, this apparent difference disappeared after correction for the influence of age.
Although antibiotics were introduced near the end of World War II, their use did not appear to result in a lower infection rate despite early optimism. In fact, some careful observers of surgical practice published articles citing a higher SSI rate with the use of antibiotics than without. This paradox was undoubtedly due to the ineffective (usually postsurgery) use of antibiotics in patients recognized by their surgeons to be at high risk of infection. The effective use of antibiotics for preventing postoperative infection was ultimately made possible by the pioneering studies of John Burke, who used an animal model to demonstrate the critical importance of the timing of prophylactic antibiotic administration [5]. He showed via a guinea pig model that the appropriate antibiotics given before bacterial contamination could cause a significant reduction in the risk of infection while the same antibiotic, given after bacterial contamination, was much less effective. This information was translated into trials demonstrating clinically and statistically significant effects in human patients undergoing scheduled operative procedures, first by Bernard and Cole [6] and then by Polk and Lopez-Mayor [7] in the 1960s. Work on prophylactic antibiotics since that time has focused on defining those procedures and circumstances most likely to benefit from the use of prophylactic antibiotics and on examining the relative efficacy of different drugs and different routes and regimens of administration (see Chapter 13).
As improvements in anesthetic care and understanding of surgical physiology permitted more aggressive and widespread surgical intervention during the second half of the 20th century, the importance of surveillance for infectious complications became more evident. One result of the NRC study cited earlier was the observation of widespread differences in SSI rates among the participating hospitals for similar classes of wounds. This was probably one of the inspirations for the careful analysis of additional factors influencing infection risk that was carried out with that study. Data of this sort encouraged the systematic collection of information regarding postoperative infection rates and factors known to influence these rates.
In the 1970s, the Centers for Disease Control and Prevention (CDC) began the National Nosocomial Infections Surveillance (NNIS) system [8]. Although it included all healthcare-associated infections (HAIs), one component emphasized from the beginning the collection of data on postoperative infections. Data from the NNIS system provide a rich source of information about the relative occurrence of infections at all sites in hospitalized surgical patients [9]. Also, in the 1970s, surgical groups' reports of surveillance of large numbers of incisions validated the relationship between wound class and different risks of infection as well as the beneficial effect of reporting SSI rate data to the operating surgeons [10].
Biology of Surgical Site Infections
SSIs are caused by bacteria, and, in the absence of bacteria, they do not occur. However, surgeons have known for years that many other factors also influence the risk of infection. Burke demonstrated in 1963 that all (50/50) clean surgical incisions contain bacteria at the end of an operation, but only a small number (4% in that report) become infected [11]. An animal study of the relationship between bacterial inoculum and SSI risk showed an increasing danger of infection with increasing numbers of bacteria. This risk was described by a typical sigmoid, biologic curve when inoculum size was graphed against SSI incidence. However, there was no inoculum in that study of 1,028 incisions that resulted in either a zero or a 100% risk of infection [12]. The authors concluded that the development of infection in a surgical incision is “dependent on many factors other than the presence of bacteria.” They further predicted that reductions in the incidence of postoperative infection could be achieved both by using techniques to reduce the numbers of bacteria that gain access to surgical wounds and by focusing on methods to increase the efficiency of host defenses in resisting those bacteria that do gain access to the wound. Modern surgical surveillance and surgical infection control must acknowledge both of these areas to achieve the goals of minimum postoperative infection rates.
Surgical Surveillance and Classification of Surgical Wounds
As indicated earlier, the oldest and best-established definitions of surgical wound classes originated with the NRC study of the efficacy of ultraviolet light for reducing wound infections. That study placed all wounds into one of five classes [2]:
P.585
or biliary tract was considered clean if the urine and/or bile were sterile.
All reports since the original NRC report have condensed this system into four groups, combining refined-clean and other (clean) into the one category of clean. Subsequent reports indicate a general consistency of the trends toward decreased overall SSI rates that is most marked in the contaminated and dirty classes of wounds (Table 35-1) [2,3,4,10,13]. These rates could have been influenced by a variety of factors, including a better understanding of the effective use of prophylactic antibiotics and of the bacteriology of dirty operative procedures and a reduction in the practice of closing the skin in dirty procedures.
|
TABLE 35-1 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Since the NRC study, much effort has focused on understanding which factors other than wound class affect the SSI risk. This trend began with the original analysis of additional risk factors performed with the NRC study. The earliest efforts to control SSI focused on lowering infection rates for clean wounds because these wounds should theoretically have a zero SSI rate if all bacteria could be eliminated from the wound. Thus, efforts focused on aseptic technique for prevention of SSI. Subsequent work found that even clean wounds become contaminated with some bacteria, and evaluation of historical data (Table 35-1) discovered a potential for reducing SSI rates even in high-risk wounds. This provided an incentive to understanding the underlying SSI risk in order to sensibly compare inter- or intrafacility SSI rates.
Another major effort in this area came from the Study of Efficacy of Nosocomial Infection Control (SENIC) project initiated by the CDC in 1974. The SENIC project collected data on SSIs and potential risk factors from 59,352 surgical patients admitted and operated on in 388 representative U.S. hospitals during 1975 and 1976 [3]. Multivariate analysis identified risk factors with roughly equal weight in predicting SSIs: having an abdominal operation, an operation that lasted >2 hours, a contaminated or dirty operation by the traditional NRC definitions, and ≥3 discharge diagnoses. This SENIC risk index was more discriminating than the old NRC classification of wounds (Table 35-2). The range of relative risks of SSI among clean wounds with different SENIC risk indexes is 1:14, and that of clean-contaminated wounds is 1:30; within the SENIC risk index, the range is 1:2.4 among wounds with one risk factor and <1:2 for all other risks. One weakness of the SENIC index was the employment of the number of discharge diagnoses as a factor because this number can be determined accurately only at the time of discharge.
The CDC subsequently developed a simplified risk index based on analyses of NNIS SSI data [4]. In the NNIS SSI risk index, the anesthesiologist's preoperative assessment according to the physical status index of the American Society of Anesthesiology (ASA) [14] is used instead of the number of discharge diagnoses. The ASA index assigns one point for a preoperative assessment score of 3, 4, or 5. Instead of counting any operation lasting >2 hours, a cut point was developed using the 75th percentile for operative duration for most operative procedures. A point is assigned for operative duration >75th percentile. The wound classification of contaminated or dirty is retained as in the SENIC risk index and adds one point to the risk score. The risk factor for abdominal operation is dropped. Thus, the NNIS SSI risk index has a possible score of 0 to 3.
A comparison of the predictive accuracy of the NNIS SSI risk index with the old NRC classification shows that this simpler index retains the increased accuracy and consistency within risk strata of the SENIC index while
P.586
being easier to apply (Table 35-3). One can see that the ratio of risks within single NRC wound classes range between 3.9 and 5.4 while all risks within single NNIS risk strata the ratios fall between 1.0 and 2.1. For surveillance programs with limited resources, it can be seen that surveillance of the 53% of patients with ≥1 SSI risk factors would yield data on 75% of all SSIs, thus increasing the efficiency of surveillance efforts [4].
|
Table 35.2 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
TABLE 35.3 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Despite its advantages over the NRC wound classification system as a sole method of estimating SSI risk, the NNIS index, as with all indexes, cannot predict the outcomes for individual patients. In addition, the NNIS index lacks predictive power for certain highly standardized procedures, such as coronary artery bypass grafting [15], cesarean section [16], and craniotomy [17]. Although the NNIS index can accurately distinguish the risk of procedures from different categories of operative procedures, it does a poor job of distinguishing higher- and lower-risk procedures among all patients undergoing the same procedure. In these instances, different risk factors specific to the procedure and to the population become more important. Another potential problem with the NNIS system is inconsistency in assignment of ASA scores [18]. A comparison of the sensitivity and specificity of ASA scores compared with the presence of ≥3 discharge diagnoses would be of interest. This comparison could probably be carried out on the original data sets used in the studies by Haley et al. [3] and Culver et al. [4].
Host Factors That Influence Infection Risk
Many individual host factors influence SSI risk. Several were listed in the discussion of the NRC study earlier in this chapter. Most have been determined in studies like the NRC study in which patients undergoing one procedure or a variety of procedures and stratified by other known risk factors and/or by multivariate analysis have been followed to determine outcomes and the association of postulated risk factors with those outcomes. In most instances, the precise mechanism of action that links the risk factor and the infectious outcome are not known, although plausible explanations often have been provided based on logical reasoning but not on proof. Thus, the increased
P.587
SSI rates observed with advanced age, extreme obesity, weight loss, hypoalbuminemia, or diabetes mellitus have been attributed to nonspecific defects in host defenses. The increased SSI risk observed in patients with anergy is not easily related to other measurable immune functions.
Patients who have an active infection at another body site are at increased risk of postoperative SSI [2]. This finding could relate to the increased risk that significant numbers of bacteria will gain access to the wound during the procedure or to inoculation of the surgical wound by bacteremia [19]. Data from human wounds suggest that the risk of postoperative infection is very great whenever the wound inoculum is >105 bacteria [20]. Although lymphatics have been suspected as a route of infection in patients with distal infections, evidence is lacking [21]. Patients who have been shaved at the surgical site before the time of operation have a higher risk of infection, which is thought to result from the tiny nicks and cuts caused by the razor and subsequent bacterial proliferation and inflammation in those injuries [22]. In vascular surgery, similar operations have a higher risk of postoperative infection in the groin region than in the arm or neck [23]. This could stem from local vascularity, local differences in bacterial number and type, or both.
Intraoperative Events That Influence Infection Risk
SSI risk is correlated with several factors that can be measured in the operating room. Other than wound class, one of the most consistently reported factors is the duration of the operative procedure. The precise connection between duration and SSI risk is not known. It is plausible that a prolonged operation results in more desiccation of tissues, potential for hypothermia [24] of the patient, and increased exposure of the wound to bacteria. It also is possible, however, that a longer operative duration is a marker for other, unmeasured factors, such as the underlying difficulty of the procedure, more scarring, larger tumor, patient obesity, or difficulty in exposure. No trial has been, or will be, conducted in which the same surgeon in performing procedures of equal severity deliberately varies the duration of the procedure. An operation that is rushed could heighten the risk of intraoperative contamination or of imperfect hemostasis with subsequent increased SSI risk. Operations should not be prolonged unnecessarily, but emphasis on the speed of operation can be misleading.
Operative technique and skill are widely believed to influence the SSI risk and other operative complications, but direct evidence is wanting. However, it can be shown that gross contamination of the operative field [25], poor hemostasis in the wound [7], and the need for all homologous blood transfusions [26] are associated with a higher incidence of postoperative infection. Surgeons commonly believe that leaving “dead space” or fluid in a wound increases the SSI risk. However, animal models of SSI demonstrate that placing sutures to eliminate dead space [27] or placing drains to close wounds and evacuate fluid [28] amplifies the SSI risk. Few prospective, controlled trials of this type are available for clinical surgery [29]; however, most observational studies demonstrate a greatly increased SSI risk with the use of drains [30,31].
Anything that supports the access of bacteria to the operative wound will increase the SSI risk. As discussed earlier, in intestinal operations, gross spill is associated with a higher SSI rate. In clean operations, a lapse in technique or in the usual protective devices is associated with increased SSI risk. In particular, glove punctures [32,33] or other episodes of contamination of the sterile field [33] raise the SSI rate. While few data directly confirm the value of hand scrubbing by the surgical team, the practice is firmly entrenched and presumably provides some protection against serious bacterial contamination when glove punctures occur. One observational study provides support for the importance of good hand-hygiene techniques [34]. The efficiency of the surgical drapes and gowns in preventing bacterial access to the surgical field also influences SSI risk. Impervious synthetic drapes result in a lower SSI rates in clean wounds when compared with those made of more loosely woven materials [35]. While most bacteria that cause SSIs come from the patient's endogenous flora and a smaller number come by direct contact from the hands and torn gloves of the operating team, occasional SSI epidemics can be traced to a member of the operating team, including a circulating nurse or anesthesiologist who does not have direct contact with the wound [36,37]. Such SSIs can occur without observed breaches of aseptic practice.
Preparation of the Patient and Operative Site
Despite much study, some common practices are of unknown efficacy for preventing SSIs. Different reports indicate reduction in SSIs or no effect from preoperatively bathing the patient with antiseptic soaps [38]. If practical, the lowest SSI rate for clean wounds is associated with no hair removal at the operative site. If hair removal is necessary, depilatory is preferable to clipping, which is preferable to shaving. Any hair removal should be done as close to the time of incision as possible [22]. Several agents are suitable for antiseptic preparation of the operative site. They include chlorhexidine preparations, various iodophor compounds, or tincture of iodine. To be effective, the antiseptic must be allowed to dry naturally before removal. Most operative sites are prepared first by the use of an antiseptic soap followed by an antiseptic preparation without detergent. Various commercial systems that employ foam or adherent plastic drapes with an antiseptic contained within the adhesive also seem to produce acceptable reduction in resident bacteria at the operative site while enhancing convenience or saving time in the operating room.
P.588
Antimicrobial Prophylaxis
Since publication of the first articles [6,7] that cited the effectiveness of perioperative prophylactic antimicrobial administration, literally thousands of other articles covering this topic have been published (see Chapter 13). Recommendations from a recent review sponsored by the Infectious Diseases Society of America and endorsed by the Surgical Infection Society (SIS) are summarized in Tables 35-4, 35-5, and 35-6 [39]. Each recommendation given in these tables is classified according to the strength of the recommendation for or against a particular use and according to the quality of the evidence on which the recommendation is based. These categories are adapted from those published in an article by McGowan et al. [40] and Gross et al. [41].
For most questions, regarding antimicrobial prophylaxis there is a consensus about general principles. The most contentious areas remaining relate to the use of prophylaxis for some clean operative procedures, the specific agent used in some procedures, the duration of antimicrobial administration, and the relative merits of oral antimicrobial agents, parenteral antimicrobial agents, or both for prophylaxis in colorectal procedures. Most practitioners agree that antimicrobial prophylaxis is beneficial for procedures that involve entry into the gastrointestinal tract with resulting exposure of the surgical site to endogenous intestinal bacteria.
In gastric operations, the highly acid gastric contents keep endogenous bacterial numbers very low in patients undergoing elective peptic ulcer operations, and antimicrobial prophylaxis is not considered necessary. With the recent understanding of the role of Helicobacter pylori in the pathogenesis of peptic ulcer disease, elective operations for this condition have essentially ceased. Procedures on the stomach for cancer, gastric ulcer, bleeding, obstruction, or perforation are considered high risk because of the higher bacterial densities encountered and the increased risk of postoperative infection; thus, prophylaxis is recommended for these types of surgery. Prophylaxis also is recommended for gastric operations on morbidly obese patients [42,43,44,45,46,47].
The biliary tract is sterile in healthy persons, and colonization rates are low during elective operations for symptomatic stone disease. Higher rates of colonization and of postoperative infection are encountered in patients >60 years old or who have common duct stones, bile duct obstruction, recent episodes of acute cholecystitis, or previous operations on the biliary tract. Antimicrobial prophylaxis is recommended for patients in these high-risk categories [39,42,43,44,45,46,47].
Elective colon and rectal procedures are followed by very high SSI rates in the absence of antimicrobial prophylaxis, and such prophylaxis is widely practiced [46,48]. For procedures other than colorectal operations, parenteral administration of the antimicrobial agent is standard. For colorectal procedures, the opportunity exists to use oral (luminal) agents to lower the endogenous bacterial load before operation. Both routes of antimicrobial administration have been found to reduce SSIs when compared with those of a placebo, but the additional benefit of using both together has not been firmly established [48]. Nevertheless, the most common practice in the United States is to achieve mechanical cleansing of the bowel combined with oral administration of antimicrobial agents designed to reduce the luminal bacterial load on the day before the scheduled procedure and to administer parenteral agents in the operating room immediately before the operation [48,49]. The oral regimen that has the best support is the administration of 1 gram (g) each of neomycin and erythromycin base at 19, 18, and 9 hours before the scheduled time for the colon procedure [50].
Some procedures do not enter the gastrointestinal tract but nevertheless have a high rate of postoperative infection without prophylaxis. They include lower-extremity vascular procedures, hysterectomy, primary cesarean section, and craniotomy. Some other procedures do not have excessive SSI rates, but any SSIs that do occur have devastating consequences. These operations include joint replacement or placement of other prosthetic devices, cardiac procedures, and aortic graft placements (see Chapters 38 and 39). These procedures are widely regarded as benefiting from prophylactic antimicrobial administration [39,42,43,44,45,46,47].
The use of prophylactic antimicrobial agents for clean operations in which the SSI risk is relatively low and the consequences of infection are considered mild is controversial [43,51,52]. When SSI rates are low and the consequences small, the use of antimicrobial agents could expose patients to a drug to prevent a single infection, which could predispose to the development of bacterial resistance and/or an increase in adverse drug reactions in the population being treated. One proposal has been to limit the use of prophylactic antimicrobial agents to those clean procedures that can be demonstrated to carry a high SSI risk using one of the risk indexes discussed previously [46].
Many agents have been found to be effective for perioperative prophylaxis. In recent years, most new antimicrobial agents with any potential for surgical use have been licensed with one or more prophylaxis indications. The primary requirement for a prophylactic antimicrobial agent is that it be active against the pathogens known to be present at the operative site and those typically recovered from SSIs. The agent most commonly recommended for procedures that do not involve the distal ileum, colon, rectum, or appendix, and that, therefore, do not entail much risk of exposure to colonic anaerobes, is cefazolin [39,42,43,44,45,46,47]. Procedures that do involve these sites require an antimicrobial agent with activity against Bacteroides fragilis and other colonic anaerobes and the Enterobacteriaceae [48]. Cefoxitin and cefotetan are the two most commonly recommended agents. Newer, so-called advanced-generation agents have
P.589
P.590
not been confirmed to be superior to cefazolin, cefoxitin, or cefotetan for the prophylactic indications for which these three drugs have been recommended [47]. Regimens with specific activity against Enterococcus spp. have not been shown to achieve superior results in colorectal procedures. Such regimens (ampicillin, amoxicillin, and vancomycin combined with gentamicin) are recommended for prophylaxis against bacterial endocarditis when gastrointestinal or genitourinary procedures are performed on patients with high-risk cardiac conditions [53].
|
TABLE 35.4 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
TABLE 35.5 |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
TABLE 35-6 |
||||||||||||||||||||||||
|
||||||||||||||||||||||||
The timing of antimicrobial administration for prophylaxis is important (see Chapter 13). For maximum effectiveness, the agent must have high concentrations in blood and body fluids before an incision is made. The original successful reports of prophylaxis specified that drugs were administered to patients “on call” to the operating room [7]. Before then, the practice had been to begin antibiotic therapy in the recovery room after the completion of the procedure. This was ineffective. Recent reports have confirmed the importance of administering prophylactic agents in the immediate preoperative period, usually within 30–60 minutes of the incision (unless vancomycin or quinolones are used, in which instance administration should occur 1–2 hours before the incision) [54,55]. Unfortunately, postoperative initiation of “prophylaxis” is still relatively common [54,56,57].
The necessary duration of antimicrobial prophylaxis has not been established, although analysis of data from some published reports suggests that antimicrobial activity should be present in blood and wound fluids at the time of closure for maximal effectiveness. Most early trials employed administration on a schedule that gave the third and last dose 12 hours after the first preoperative dose. Numerous reports indicate that longer durations and prolonged administration of prophylactic agents are common in clinical practice [54,56,57]. Although a single report suggests better results with longer use of prophylactic therapy in certain high-risk patients undergoing peripheral vascular procedures [58], most published articles support a short duration of antimicrobial prophylaxis [47,59,60]. Another area of controversy is the prevention of SSI after traumatic injuries that expose the patient to
P.591
bacterial contamination before antimicrobial treatment can be initiated. Because prophylactic agents cannot be administered before contamination, some clinicians have termed their use therapeutic rather than prophylactic. However, prophylaxis can be achieved for the operative wounds created in treating the injuries. A number of recent studies have established the fact that even in this context, a short duration of antibiotic treatment is as effective as a prolonged one and probably is preferable [60,61].
Selective Digestive Decontamination
More than 10 years ago, Stoutenbeek et al. proposed a method of selective digestive decontamination (SDD) to lower the infection rate in posttraumatic surgical patients in the intensive care unit (ICU) [62]. The concept holds that potentially pathogenic bacilli reside in an intestinal reservoir and serve as the source for many HAIs. The proposed regimen—tobramycin, polymyxin B, and amphotericin B by nasogastric tube and oral paste for the duration of ICU stay and parenteral cefotaxime for the first 5 days—is supposed to suppress potentially pathogenic microorganisms in the gastrointestinal tract while preserving the useful anaerobic flora that promote colonization resistance. Despite years of study and numerous publications, the concept is still the subject of much controversy. It has many proponents in Europe but has failed to garner much enthusiasm or be put into wide practice in the United States except among some solid-organ transplant groups [63]. While many articles report a reduction in nosocomial lower respiratory tract infections, few cite any important differences in mortality rates or duration of ICU stay, and the suspicion arises that what has changed is the diagnosis of pneumonia, a notoriously difficult task in ICU patients (see Chapters 24a, 24b, 31). A number of the articles written about SDD report the early emergence of
P.592
resistant gram-positive cocci [64,65]. Some recent reviews make clear that the role of SDD is still very controversial [65,66,67].
Systemic Surveillance of Postoperative Infection
The basic elements of a successful infection surveillance program include a definition of specific types of postoperative infections, a method for screening patients at risk, and a reliable means of recording and retrieving information (see Chapter 6). The observer who records the presence or absence of infection (infection control professional [ICP] or hospital epidemiologist) should have no conflict of interest in performing these duties. The surgical staff should be aware of methodology for screening and recording SSIs and should accept the definition criteria. It is useful for the observer to make rounds with several surgeons to ensure agreement on the use of definitions. The general criteria put forth by the CDC [68] have recently been updated (Table 35-7). A common definition error is to equate infection with recovery of a bacteria from the wound site. Bacteriologic findings do not distinguish between colonization or infection.
Surveillance objectives should be defined. All SSIs are not necessarily equally important to identify. Highest priority is given to detecting SSIs that lead to death, reoperation, increased intensity of services, or special diagnostic or therapeutic measures (including parenteral administration of antimicrobials, prolongation of hospital stay, or hospital readmission). Lower priority is assigned to finding superficial SSIs that do not lead to the outcomes just cited and that are readily treated on an ambulatory basis without a major increase in therapeutic costs or delay in the patient's return to regular activities. SSIs (especially deep ones complicating operations involving cardiac, vascular, and neurologic structures; bones; joints; stomach; bowel; or rectum) are more likely to compromise a patient's well-being than are similar SSIs complicating hernia repair or uterus, gall bladder, or thyroid operations.
Postoperative infections beyond the operative site may bring about considerable morbidity and should be included in routine systematic surveillance. This is particularly true for pneumonia (see Chapter 31). In special-care surgical units, surveillance of various infections caused by multidrug-resistant bacteria can be especially useful in providing information to surgeons about possible inappropriate antibiotic use and could indicate personnel or environmental transmission of bacteria. Hospitalwide outbreaks of infections due to methicillin-resistant S. aureus or vancomycin-resistant enterococci and clusters of diarrhea or colitis due to Clostridium difficile can manifest first among postoperative patients (seeChapter 33). Surveillance activities should identify such infections.
Hospital episodes of SSIs and postoperative episodes of bacteremia, pneumonia, and symptomatic urinary tract infection should be identified for the three categories of elective clean, clean-contaminated, and contaminated operations. Whereas patients having dirty operations could acquire a new infection as a complication of the operative procedure, often it is difficult to separate this event from extension of the infection already present at the time of operation. As the surveillance findings from patients undergoing specific types or groups of operations become large enough for meaningful interpretation, SSI rates for various procedures can be established as a baseline and to reflect the specific populations of patients being treated. After a database has been collected, cluster definitions and threshold levels for outbreak investigation could be defined.
The relative sensitivities of various methods for finding HAIs have been reported [69] (see Chapter 6). It may not be possible because of methodologic and interobserver differences and those in groups of patients to directly compare HAI rates complicating specific operations done at different hospitals. At the Florida Consortium for Infection Control, an organization of nonteaching community hospitals, a retrospective medical audit was useful in assessing the accuracy of ICPs in reporting all categories of postoperative infections. The expected ICP sensitivity for detecting SSIs at the operative site is ≥80%; the specificity is ≥97%. At the Minneapolis VA Medical Center, the ICP uses the floor nurses who see patients' wounds as part of daily care as sensitive indicators [70]. The nurses have been trained to recognize and to report to the ICPs all clinically suspicious wounds. ICPs personally examine such wounds and determine the wound status. In addition, every nurse on a surgical ward is authorized to send wound fluid for culture without a specific order.
A positive culture is not by itself a sign of infection. Instead, the fact that a culture was sent alerts the ICP that the wound was regarded as suspicious and prompts the ICP to inspect the wound personally. This system allows the ICP to focus his or her efforts on only the most suspicious wounds and provides a complete microbiologic record for those wounds that are determined to be infected [70]. Cardo et al. have reported on the accuracy of a similar system and found that sensitivity varied between 80–90%, depending on the degree of the ICP's experience, and that specificity was >99% [71].
The surveillance methods described here are all labor intensive. Moreover, the ICP's learning period can vary from several months to several years [71,72], depending on the intensity of training. Once a satisfactory level of surveillance sensitivity is achieved, it can be maintained only in direct proportion to the time available for the ICP to conduct surveillance. Yokoe and Platt have outlined a novel screening method in which antibiotic treatment of patients serves as a marker of those with possible SSIs [73]. This method has some promise to be labor efficient.
P.593
P.594
|
TABLE 35-7 |
||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||
Any SSI rate determined in such a system is, of course, a minimum estimate of the infection rate because wounds that escape surveillance or that manifest only after discharge will not be counted. A recent consensus paper by the joint Surgical Wound Infection Task Force representing the Society for Hospital Epidemiology of America, Association of Professionals in Infection Control, the CDC, and the SIS has recommended that some form of postdischarge surveillance should be undertaken [74]. Tools to identify postdischarge infections include questionnaires sent to surgeons or to patients, telephone follow-up with surgeons or patients, and systems to detect readmission of patients to hospitals and postoperative antibiotic prescriptions. The best method for accomplishing such surveillance is not known, and some systems appear to be both expensive and insensitive [75]. In addition, postdischarge surveillance can be more useful for some procedures than others. No consensus has been reached on what method of postdischarge surveillance and for what procedures such methods should be used.
Another area that has received very little attention to date is the surveillance of outpatient operative procedures. The NNIS system specifically tracks only inpatients [8,68]. An increasing number of operative procedures are being performed on patients who are not admitted to a hospital, ≥50% in many hospitals (see Chapter 28). Preliminary information suggests that SSIs in these patients are less common than in patients having similar procedures as inpatients [76,77]. However, very little systematic information is available. As more and more procedures are performed on an outpatient basis, the usefulness and validity of postdischarge surveillance for SSIs in this population will continue to be debated.
A surveillance worksheet indicating the criteria used for SSI definitions is useful to record important details about an infected patient. Only the minimum information necessary for further tabulation to identify clusters or outbreaks should be included. In general, patients' demographic characteristics and details of operations are readily available on-line or in the medical record and need not be routinely recorded. On infrequent occasions when SSI clusters are found, additional information about host risk factors, technical aspects of surgical procedures, and environmental details concerning the patients' care can be important for stratification during analysis. Computer analysis of surveillance data is helpful only to the extent that the data output is consonant with the previously determined objectives (see Chapter 8).
Analyses of Crude Data: Investigations of Outbreaks and Clusters
In the workup of an apparent excess frequency of SSIs, acquisition of infection can be evaluated in relation to the sequence of surgical care: preoperative (therapeutic and host factors), intraoperative, and postoperative. Identifying a specific operation as the point of selection rather than as an operator, a wide range of possible contributions to individual infection risks is more likely to be fully explored. Operations should be surveyed for postoperative infections at operative and nonoperative sites.
It is especially important to recognize the potentially pejorative implications of incorrectly interpreting crude data sets of postoperative infections. Dissemination of crude data to an uncritical audience has the power to damage a surgeon or an institution. Access to crude data sets should be restricted to those who have an actual need to know about interval findings of work in progress. The limitations of crude findings of postoperative infections should be identified for those who review them. There also should be constant concern about disseminating interpretations of a single set of data indicating highly unusual findings because of possible sampling or other biases and for type 1 error due to inadequate sample size (see Chapter 7). Failure to identify potential limitations in interpreting crude data and possible sampling errors in small clusters can result in a loss of credibility of the surveillance program, create a climate of suspicion and animosity among the surgical staff, and jeopardize the success of any future efforts. Thus, sound epidemiologic methods and strict confidentiality of records are central to an effective program. However, the recent passage of legislation requiring mandatory reporting of hospital-specific HAI rates, although publicized as beneficial, could lead to efforts to minimize SSI rates and not ultimately lead to reduction of HAIs.
SSI cluster can be sought by grouping infected patients according to characteristics such as time, place, person, infecting organisms, surgeons, types of operations, preoperative host and surgical risk factors, use of various devices, characteristics of critical care, preoperative duration of hospitalization, and postoperative intervals until onset of infection (seeChapter 6). In stratification, the analyst seeks to identify possible commonality in SSI sources. For example, a lengthy preoperative period of hospitalization could indicate patients with unusual or complex manifestations of disease. Relatively long periods from operation to SSI onset can suggest that a postoperative event has played a role. In the latter instance, time periods in >10 to 12 days raise the possibility that postoperative factors (including hematoma, seroma, remote site infection, prolonged or unnecessary use of surgical drains, and postoperative manipulation of the wound) aided in the initiation of infection. An open conduit at the surgical site (e.g., Penrose drain) also could be a factor in permitting infection.
An increase in the incidence of postoperative infection could result from a change in risk characteristics of the patients with the introduction of a population having an increased susceptibility to infection. In small SSI clusters, chance alone can bring about a concatenation of unrelated
P.595
events, giving the false appearance of a group of SSIs linked by common causation. Alternatively, a small SSI cluster in high-risk patients could indicate a much larger outbreak with a wide array of disease manifestations (see Chapter 6). A systematic search for unreported SSIs, including those arising after hospital discharge, usually is necessary to determine the actual size of such clusters. Case-control analysis, stratified as necessary, is useful when new equipment, procedures, or processes have been introduced.
Surgeon-Specific Surveillance
Following the outline of the 1964 NRC study [2], Cruse and Foord [10] devised a program of concurrent observation of surgical wounds in patients undergoing clean operations to provide individual surgeons with SSI rates. This study and others [13] have demonstrated that providing SSI rates to the surgical staff results in a decrease in SSIs. These findings are consonant with the results of the SENIC study [69], in which reduction of SSIs was found to be associated with a strong infection control program that included an effective hospital epidemiologist and a system for reporting specific SSI rates to the specific surgeon. Surgeon-specific SSI rates can be readily determined. Aggregated SSI rates for different classes of operations can be calculated for individual surgeons from NNIS SSI risk index data and compared with a standardized infection ratio [78].
The goal of surgeon-specific surveillance is to lower the SSI rate by making individual surgeons aware of excesses in SSI rates and thereby promote adherence to accepted principles of operative care. Implicit in this approach is the assumption that SSIs can be entirely attributed to flaws in surgical techniques or judgment (including incorrect use of antibiotic prophylaxis). Also implicit is the assumption that any reduction in SSI rates following the provision of information regarding excess rates to an individual surgeon results from improvement in faulty surgical technique, protocols, and/or judgment. Massanari has pointed out some concerns for small sample sizes, validity of methods for risk adjustment, and reliability of data collection methods in this context [79].
Nonetheless, an actual problem in surgical technique could exist for one or several surgeons. This is a thorny issue. There is no methodology to evaluate operative technique as a surveillance activity. Moreover, providing some surgeons with surgeon-specific data could not bring the desired results. In some instances, surgeons in community hospitals with a high rate of postoperative infection have simply moved their practices to other hospitals in which surveillance is less intense.
Lee has suggested that rather than focusing too much attention on absolute SSI rates, the hospital epidemiologist and surgeon should concentrate on classifying each SSI into the dichotomous classes of potentially avoidable and apparently unavoidable infections [70]. A potentially avoidable SSI is an infection that occurs under circumstances in which any indicated infection-reducing adjuncts were omitted. Such adjuncts would include appropriate use of prophylactic antibiotics (agent, timing, and discontinuing), avoidance of razor shaving on the day before operation, appropriate skin preparation at the operative site, avoiding elective operations in the presence of distant site and active infections, and so on. An apparently unavoidable SSI is one that occurs despite the application of all known appropriate infection-reducing measures. The goal of an SSI surveillance program should be zero avoidable SSIs [70]. When a particular infection-reducing strategy is well accepted but not always reliably applied, such as the indicated use of perioperative prophylactic antibiotics, a quality assurance program designed to ensure the reliable application of this method could have a greater effect on SSI rates than could feedback on SSI rates alone [39].
Viral Infections
Studies of patients receiving transplants have confirmed that viral infection is a surgical complication in this context [80] (see Chapter 45). The principal agents are transferred by human tissue organs and blood. Cytomegalovirus, herpesvirus, hepatitis B, hepatitis C, and varicella zoster are the viruses most frequently identified (see Chapter 42). Human immunodeficiency virus is the one most feared (see Chapter 43). Routine methods for surveillance of viral postoperative infections need to be developed.
Fungal Infections
Candida spp. are gaining prominence in consideration as causes of SSI. This is particularly true in the surgical treatment of transplant patients, burn patients, and patients with prosthetic devices (see Chapter 43). Intravenous catheter and urine cultures can yield fungi before systemic involvement, and it is likely that patients are infected from such sources by hematogenous spread. Improved methods to identify and control such infections will become more important as technical advances in surgery lead to treatment of more immunocompromised subjects.
The Costs of Surgical Infections
Postoperative infections in surgical patients can prolong the length of hospitalization for substantial periods, depending on the type of operation, and thereby greatly increase the cost of their care. Cardiothoracic, orthopedic, and gastrointestinal operations are likely to be especially costly in this regard as the result of both pulmonary and operative site infections [81,82]. In addition to the higher direct costs of care, indirect costs should be considered in calculating
P.596
the consequences of postoperative infection. These costs include the time the patient loses from gainful employment and the possible medicolegal actions that the patient could take against a hospital or the surgical staff (see Chapter 17).
New Directions in Preventing Postoperative Infection
As the incidence of SSIs progressively declines, new thinking is required to define the next step in improving the care of patients in relation to controlling postoperative infection. Perioperative colonization with virulent bacteria can be a source of both endogenous infection and cross-infection. The association between nasal carriage of S. aureus (or MRSA) and postoperative infection is well known [83]. Some evidence exists that topical nasal treatment with mupirocin can reduce endogenous SSIs with S. aureus [84,85,86] and minimize S. aureus cross-infection in a surgical ICU (see Chapter 40). Both nasal colonization with S. aureus [85] and lower respiratory tract colonization with Haemophilus spp. [87] appear to be sources of postoperative pulmonary infection. How to identify and reduce the latter, especially in the preoperative treatment of the patient with traumatic injury, remains to be established. Before, during, and after surgery, a number of additional potential measures for controlling SSIs [88,89,90,91,92,93], such as intravenous administration of immunoglobulins, could lessen the frequency of postoperative acquisition of bloodstream infection or pneumonia. Aggressive and consistent infection control activities will be important in further reducing the incidence of postoperative infections.
Recently, the Center for Medicare and Medicaid Services (CMS) initiated a national SSI prevention initiative [94,95,96]. This effort focused on the appropriate use of prophylactic antibiotics in selected surgical patients. The major initial focus was on evaluating the appropriateness of antibiotic prophylaxis; this study documented that the correct drug was given at the correct time and stopped correctly in only 38% of patients [97]. The Surgical Infection Prevention Project (SIPP) has evolved to the Surgical Care Improvement Project (SCIP). In the second phase, prospective improvements in prophylactic antibiotic use, glucose control, and normothermia in selected surgical patients will be initiated. These projects can be the forerunners to CMS's paying for performance; those who have high compliance with performance indicators will be paid more and those who have lower rates will be paid less.
References
P.597
P.598