Chapter 64 Antimicrobial Prophylaxis in Ophthalmology BRUCE D. GAYNOR and MICHAEL B. STARR Table Of Contents |
Prophylaxis of ocular infections dates from at least the 19th century, with
development of the Credé technique of silver nitrate instillation
to prevent gonococcal ophthalmia in the newborn.1 Today, opportunities abound for ocular infection prevention, using antibiotics, antiviral
agents, antiseptics, antitoxins, and immunization. The
clinical settings include not only the prevention of postoperative
infections but also posttraumatic intraocular and eye wall infections
and tetanus; infections after removal of suture material and implantation
of human donor tissues such as cornea, sclera, and fascia lata; and
recurrent herpes simplex keratitis and cytomegalovirus (CMV) retinitis. Rational prophylaxis of ocular infections is a challenging task because it requires consideration of the pharmacokinetics of drug delivery, optimal in vitro activity against the most likely offending pathogens, and demonstrated clinical efficacy in in vivo applications. It also encompasses more thorny aspects, such as the problems of promoting antibiotic resistance to potent antibiotics that are effective in treating the most antibiotic-resistant bacteria, and the ethics of preventing infection in an individual patient, possibly at the expense of the future welfare of the community of all patients. Indiscriminate and unthinking use of prophylaxis is potentially harmful to all and must be counteracted by well-founded and deliberate prophylaxis. This is all the more challenging in ophthalmology, in which there are fewer clinical studies to support and direct prophylactic regimens than in other medical and surgical specialties. This chapter discusses the scientific basis and specific recommendations for various prophylactic regimens in ophthalmology. |
PROPHYLAXIS OF POSTOPERATIVE INFECTIONS |
NONOPHTHALMIC LITERATURE Perhaps the two most important factors responsible for the success of modern surgery are the use of aseptic technique and prophylactic antibiotics. The effectiveness of antibiotics during surgery was first established in an experimental animal model in the early 1960s.2 Many clinical trials have proved the effectiveness of prophylaxis; perioperative antibiotic use has become a standard of care for most major surgical procedures. Despite this, there are at least 920,000 postoperative wound infections in the 23 million patients who undergo surgery annually in the United States.3 The exact timing, choice of antibiotic, and duration of therapy are constantly being updated. Although there are no definitive studies in the ophthalmic literature, the nonophthalmic surgical experience provides a framework for the development of a treatment regimen for ocular surgery. The use of antibiotics in contaminated or emergent settings, such as perforated bowel or infected biliary or urinary tracts, is considered to be therapeutic and therefore is not included in this discussion of prophylaxis. The indications for prophylaxis include procedures wherein the risk of infection is high or the consequences of infection are particularly deleterious, even if the risk of infection is low.4 For about two decades, it has been firmly established that prophylaxis is beneficial in “clean-contaminated” procedures such as bowel surgery, in “clean” procedures involving the insertion of a prosthetic device such as total hip replacement, and in major “clean” procedures such as coronary bypass.5–7 More recently, indications for prophylaxis have extended to include relatively minor “clean” procedures such as mastectomy or herniorrhaphy.8 The success of antibiotic prophylaxis in surgery correlates directly with the susceptibility of the infecting organism in vitro to the antibiotic used.9 Because of their excellent spectrum of activity, the low incidence of allergy, and proved efficacy in clinical trials, the cephalosporins, especially cefazolin, are used most widely in the world for prophylaxis.10 Emerging pathogens with greater resistance may require rethinking of this approach in the future. Historically, one of the most controversial questions regarding prophylaxis is the timing and duration of therapy. It is generally thought that high blood levels of antibiotic should be maintained throughout the surgical procedure and that a long course is no more beneficial than a short course in the postoperative period.11–13 In a study of 2847 patients undergoing elective clean or cleancontaminated surgical procedures, it was found that preoperative (during the 2 hours before incision) antibiotic administration was responsible for a significant reduction in wound infection, compared with early (2 to 24 hours before incision); perioperative (during the 3 hours after the incision); or postoperative (between 3 and 24 hours after the incision) administration.14 In this study, there were 346 different types of surgery performed; the three most common were total abdominal hysterectomy (10%), cholecystectomy with intraoperative cholangiography (9%), and bowel resection (8%). Four intravenous antibiotics comprised 84% of all the antibiotics used: cefazolin (56%), cefonicid (12%), cefoxitin (10%), and cefamandole (6%). EMERGENCE OF RESISTANT ORGANISMS Although enterobacteria collected before the advent of modern antimicrobial agents had conjugative plasmids at about the same frequency as those isolated today, they rarely contained antibioticresistance genes, whereas plasmids isolated today often encode resistance to antibiotics.15 Bacteria are continuously developing resistance to antimicrobial agents; there are more than 100 described antimicrobe-resistance genes.16 Acquired resistance occurs by several mechanisms: alteration of the drug target, production of a detoxifying enzyme, decreased antibiotic uptake, or multidrug-resistance pumps.17 Important examples of resistance include penicillin-resistant Streptococcus pneumoniae,18 vancomycin-resistant enterococci (VRE),19 the presence of extended-spectrum β-lactamases (ESBL) in several species of gram-negative bacilli,20 and imipenem-resistant Pseudomonas aeruginosa.21 Also, multidrug-resistant Staphylococcus aureus remains a continued threat.22,23 An important example of antibiotic resistance by target alteration is penicillin-resistant Streptococcus pneumoniae. Penicillin-binding proteins (PBPs) are enzymes important in creating and maintaining the peptidoglycan component of the bacterial cell wall. Penicillins bind these enzymes and thereby interfere with cell wall synthesis. Highly resistant pneumococcal strains have PBPs with markedly decreased affinity for penicillin. Although this phenomenon was first described in 1967, widespread presence of these isolates did not occur until the 1980s.18 Roughly 25% of U.S. isolates are penicillin-resistant.24 Another example of target alteration is seen in vancomycin-resistant enterococci. Vancomycin, a glycopeptide, inhibits cell wall synthesis by forming complexes with the peptidyl-D-alanyl-D-alanine termini of peptidoglycan precursors at the cell surface. The presence of these complexes hinders prolongation of the growing peptidoglycan chain and blocks formation of interpeptide bonds. Resistant enterococci synthesize peptidoglycan precursors with the depsipeptide D-alanyl-D-lactate, which has decreased affinity for vancomycin. This resistance mechanism was first documented in France in 1988 and presents a significant clinical problem.19 Production of β-lactamases represents the second mechanism of acquired bacterial resistance, production of a detoxifying enzyme. β-Lactamases are a heterogeneous group of enzymes that hydrolyze the cyclic amide bond of the β-lactam-ring antibiotics. In the early 1980s, a new group of enzymes arose, designated the extended-spectrum β-lactamases, conferring resistance to third-generation cephalosporins in strains of Klebsiella pneumoniae, P aeruginosa, and Escherichia coli.20 Surveys of nosocomial isolates in Europe show a rising prevalence of these organisms, and multiple outbreaks in the United States have been documented.25 Imipenem-resistant P aeruginosa represents the third mechanism of resistance, decreased uptake of antibiotic. These bacteria lack an outer membrane protein, D2, through which imipenem normally traverses. The prevalence of these strains is also rising.21 Multidrug-resistance pumps are the fourth mechanism of bacterial resistance. These efflux pumps are thought to be widespread in bacteria. especially in E coli and Pseudomonas species and function by antibiotic extrusion.26–29 Some organisms display multiple types of resistance mechanisms. The development of penicillin-, methicillin-, and fluoroquinolone-resistance in S aureus is a classic example. It is notable how quickly resistance developed after introduction of these antibiotics. In the early 1940s, penicillin G was almost universally effective against S aureus. By the mid-1940s, the rapid appearance of β-lactamase provided S aureus-resistance to penicillin G; by the 1990s, more than 95% of S aureus was resistant to penicillin G, ampicillin, and the antipseudomonas penicillins.22 The emergence of methicillin-resistant S aureus (MRSA) in the 1980s due to altered PBPs spurred an increase in the use of vancomycin. Widespread use of vancomycin may have created the selective pressure favoring vancomycin-resistance in other bacterial species (e.g., enterococci). In 1996, the first documented case of S aureus with intermediate levels of resistance to vancomycin (VISA) was reported from Japan30 and isolates have been reported from the United States.31,32 Four years after the introduction of fluoroquinolones, most nosocomial MRSA was resistant to fluoroquinolones, and such resistance has been reported in almost all countries.23 Fluoroquinolone-resistance is mediated by the alteration of DNA gyrase.33 Most multidrug-resistant infections are nosocomial but these resistant organisms inevitably will spread to the community. Although resistance genes were rarely present before the use of antimicrobial agents,34 the use or overuse of these agents provided the requisite selective pressures for greater expression of these genes.35 There is evidence that subinhibitory levels of antibiotics (e.g., during tapering) provide a condition highly favorable for the selection of resistant organisms.36 Current practices by physicians include the increasing use of wider-spectrum antimicrobial agents to treat minor infections.37 Such practices must be resisted and discouraged to avoid the catastrophic potential of ever-widening antimicrobial resistance.21,38 In Finland, a decrease in the use of macrolide antibiotics has led to a decrease in the frequency of erythromycin resistance in group A streptococci.39 Widespread changes to more conservative prescription patterns would certainly slow the emergence of resistant strains but may also decrease the frequency of resistant strains that have already appeared.40 PHARMACOKINETICS Anatomic and physiologic barriers to ocular antibiotic penetration need to be considered in formulating effective prophylaxis regimens. This is equally important in interpreting the results of in vitro sensitivity testing—where sensitivity thresholds are predicated on levels of antibiotics achievable in serum or urine after systemic administration—as in determining optimal antibiotic selection, timing, dosage, and routes of administration. Cultures may be performed for a variety of ocular infections, including conjunctivitis, keratitis, and endophthalmitis; however, the results of susceptibility testing, stated simply as “resistance” or “susceptibility,” may have limited clinical relevance for the ophthalmologist.41 To use susceptibility data, one must know the usual achievable tissue levels for the antibiotic and ocular structure in question (which are rarely provided with susceptibility testing results), in addition to the in vitro susceptibility of the actual or potential infecting organisms to various antibiotics, which is often provided. The importance of both factors is underscored by the increasing use of the inhibitory quotient in addition to culture and sensitivity results in evaluating the susceptibility of potential or actual infecting organisms in ocular disease. This measure divides the achievable antibiotic level in a given ocular tissue or compartment by the minimal inhibitory concentration for a given antibiotic-bacterial isolate pair. Much has been written about the ocular tissue levels of many antibiotics, determined for various eye structures including tear film, cornea, aqueous and vitreous, when delivered by different routes (topical, subconjunctival, intracameral, oral, or parenteral).42–46 Generally, the ocular surface (and in many instances the anterior segment structures) are capable of attaining antibiotic levels that are useful in the prevention and treatment of many infections. But the more posterior tissues, particularly the vitreous, are relatively inaccessible to most antibiotics and routes of administration, except direct instillation.47 Important variables include the particular antimicrobial agent; the duration, frequency, and concentration of the antimicrobial agent administered; the presence or absence of a corneal epithelial defect (for topical and subconjunctival routes); the presence or absence of intraocular inflammation; and the presence or absence of the crystalline lens.42 Because of ease of application, limited systemic toxicity, and proved effectiveness, topically applied antibiotics have been a mainstay of ophthalmology for many years. Topical aminoglycosides have been used extensively in ocular infections and prophylaxis because of their safety and efficacy and their antipseudomonas activity.48,49 A study in rabbit and human eyes showed that with sufficiently frequent dosage, topically applied 0.3% gentamicin can achieve bacteriostatic or even bactericidal levels in the aqueous and other ocular tissues, and penetration is greater in an inflamed eye.50 Another study showed the levels attained intravitreally after topical administration in rabbit eyes to be variable (between 0 and 2.9 μg/ml).51 There is evidence that gentamicin penetration into ocular tissues is enhanced by aphakia but intravitreal levels remain less than inhibitory at best (and often unmeasurably low).52 Tobramycin has been used topically in ophthalmology for more than 20 years; there is in vitro and clinical evidence that it is more effective and is associated with fewer adverse reactions than gentamicin.48,53,54 When 0.3% tobramycin was applied topically four times daily for 2 days, however, and at 90 and 60 minutes before cataract surgery, clinically significant aqueous humor levels (13 ng/ml) were unobtainable at the start of surgery.55 Tobramycin levels in tear film are proportional to the concentration of antibiotic instilled but at higher fortified concentrations, significant decreases in rates of re-epithelialization are seen.56 Penetration of fluoroquinolones—especially ofloxacin—when applied topically, is particularly good in the anterior segment.57–59 Topically applied ofloxacin achieves significantly higher aqueous and tear film levels than topically applied tobramycin, when used at commercially available (nonfortified) concentrations.60,61 Except in the presence of a corneal defect, the penetration of cefuroxime given topically into the aqueous is poor.62 Bacitracin is a polypeptide antibiotic produced by Bacillus species and is often used topically in ophthalmology. Its penetration is poor but is enhanced when an epithelial defect is present.63 Trimethoprim has wide antibacterial coverage and is synergistic when combined with polymyxin B.64,65 Although topically applied trimethoprim/polymyxin B sulfate (Polytrim TM, Allergan Pharmaceuticals, Irvine, CA) fails to reach significant aqueous humor levels, there is evidence that it is clinically efficacious in surface ocular infections.55,66–69 Collagen shields have been advocated as drug delivery devices because of their ability to provide sustainable drug levels rapidly and without the inconvenience of frequent drops or side effects of subconjunctival injection.70 Studies on a variety of antibiotics, including tobramycin, gentamicin, vancomycin, and amphotericin B, have been done on rabbit eyes.71–75 Data from these studies show that generally, when presoaked with antimicrobial agents, collagen shields are able to attain anterior segment antibiotic levels comparable with topical drops or subconjunctival injection, but the kinetics of drug delivery seem to vary with the antibiotic in question and should be considered when shields are used in clinical applications. In one clinical series, application of gentamicin and dexamethasone by subconjunctival injection and by presoaked collagen shield was compared in patients undergoing extracapsular cataract extraction.76 No adverse effects were reported and the collagen-shield patients had less postoperative pain, less conjunctival injection, less conjunctival hemorrhage, fewer Descemet's folds, and less aqueous flare than the subconjunctival injection patients. Another potential advantage of drug delivery using collagen shields is that the shield may be used to deliver two or more medications simultaneously. Various drug combinations, such as cefazolin and tobramycin, and gentamicin and methylprednisolone, can form precipitates under certain conditions, however.77 The effect of these precipitated medications on bioavailability to ocular tissues and the potential mechanical and biochemical toxicity to ocular tissue are essentially unknown. More clinical data are needed on collagen shields, especially in humans, before recommendations regarding antibiotic prophylaxis can be made. Subconjunctival injections of antibiotics are used routinely by many surgeons for prophylaxis. Studies with a variety of antibiotics given subconjunctivally show regional differences in the concentration of the drug injected consistent with penetration of these drugs by direct diffusion rather than by tear film.78–80 Animal and human studies show adequate penetration of subconjunctival gentamicin into anterior segment tissues but poor penetration into the vitreous.79,81–83 Aqueous levels of tobramycin delivered by subconjunctival injection have been shown to be therapeutic.84–86 Subconjunctivally administered cefazolin reaches therapeutic concentrations in aqueous but vitreous penetration is poor.78,87,88 A randomized trial showed that patients given subconjunctival cefuroxime had less pain and hyperemia than those given gentamicin.89 Subconjunctivally injected cefuroxime attains clinically significant aqueous levels, whereas topically applied drops do not.62 Third-generation cephalosporins (ceftizoxime, ceftriaxone, and ceftazidime) show poor penetration into the vitreous of rabbit eyes when given 100 mg subconjunctivally; an inflamed eye facilitates drug penetration.90 Subconjunctivally injected cefotaxime reaches therapeutic levels in human vitreous only in eyes that have had previous surgery.91 Ciprofloxacin (1 mg) injected perilimbally in rabbits reaches aqueous levels of 0.887 μg/ml and 0.094 μg/ml at 1 and 10 hours, respectively.92 Penetration of antibiotics into the vitreous cavity is impeded by the blood-retina barrier; there is evidence that disruption of this barrier, such as during inflammatory conditions or trauma, facilitates penetration.42,47,50,93–98 Intravenous aminoglycosides have been advocated for trauma-associated endophthalmitis prophylaxis despite evidence of poor vitreous penetration.51,97,99,100 Because of increasing cephalosporin-resistance to coagulasenegative staphylococci, some authors have advocated empirical intravitreal and in some cases, intravenous, vancomycin for the treatment of established endophthalmitis.101–104 Therapeutic intravitreal vancomycin levels are attainable in rabbit eyes after intravenous administration only when the lens is removed, however. Ciprofloxacin has become particularly popular as an adjunct in endophthalmitis prophylaxis because of its broad spectrum of activity and good intravitreal penetration when given orally or intravenously.96,105–107 El-Baba and coworkers105 showed human intravitreal ciprofloxacin levels above the minimal inhibitory concentration for 90% of strains for Staphylococcus epidermidis, Bacillus species, and Enterobacteriaceae after a single oral dose of 750 mg; additional oral doses show even higher intravitreal levels.108 Similar therapeutic intravitreal levels were obtained in rabbits and pigs after intravenous ciprofloxacin administration.96 Three doses of 400 mg of oral ofloxacin also attained clinically significant intravitreal levels.59 The Endophthalmitis Vitrectomy Study109 showed no visual benefit to intravenous ceftazidime and amikacin in treating established postoperative (after cataract surgery or secondary lens implantation) bacterial endophthalmitis; based on this, some clinicians have argued that intravenous antibiotics have no place in prophylaxis of postoperative endophthalmitis. This may be too broad a generalization for three reasons:
The potential, however, for significant adverse effects of systemically administered antibiotics, when given to the millions of patients who undergo elective ophthalmologic surgery annually, argues strongly against routine antimicrobial prophylaxis by systemic routes, despite the limitations of extrapolating results of the Endophthalmitis Vitrectomy Study. SOURCES OF INFECTION AND THE ROLE OF PREOPERATIVE CULTURE TESTING Sources of infection in postoperative endophthalmitis include both endogenous (patients' normal resident flora on lids and conjunctiva) and exogenous (airborne contaminants, contaminated intraocular solutions, intraocular lenses, surgical instruments, and operating room personnel) sources. There are essentially four lines of reasoning that substantiate that preoperative, normal, endogenous flora is the most common source of infectious organisms in postoperative endophthalmitis: similarity regarding incidence (e.g., S epidermidis is the most common in both);111–115 phage-typing;116–118 antibiotic sensitivities;119 and molecular epidemiology using DNA probes, restriction fragment length polymorphism, and pulsed-field gel electrophoresis.120,121 The preponderance of evidence implicating endogenous flora as causative of endophthalmitis previously led to preoperative culturing and attempts to reduce or eliminate ocular flora with prophylactic antibiotics and antiseptics. Obtaining reliably reproducible culture results has proved elusive. Daily conjunctival and lid margin cultures in asymptomatic adolescents were highly variable in one study, and conjunctival cultures taken preoperatively in asymptomatic patients were reproducible 1 day later in only 65% of 65 patients without the use of antibiotics in another study.122,123 Therefore, the usefulness of these cultures in asymptomatic patients remains unclear. Even if culture results were reproducible, their usefulness would still remain doubtful. More than 95% of preoperative normal cataract patients harbor organisms that are capable of causing infection, whereas the incidence of postoperative endophthalmitis remains exceedingly low.124 It is generally thought that preoperative intraocular cultures are of dubious value and are not performed on a routine basis. The role of cultures in symptomatic patients is probably more useful, and steps should be taken to attempt sterilization of any ocular surface or lacrimal drainage system infection before performing elective surgery. Because of possible cross-contamination or generalized host defense impairment, any extraocular-site infections probably should also be treated before proceeding with elective ocular surgery.125 Finally, the clinical usefulness of in vitro sensitivity testing on ocular isolates is unproved. ANTIBIOTICS: EFFECTS ON NORMAL FLORA AND ENDOPHTHALMITIS INCIDENCE Although the results of external eye cultures are variable, there is evidence that the use of preoperative topical antibiotics decreases the amount of lid flora in asymptomatic patients and that topical antibiotics decrease postoperative bacterial counts in patients wearing patches.126–128 In a study in 1980, Fahmy129 showed that aminoglycoside drops given the night before surgery were effective in eliminating 95% of isolated strains of bacteria by the next morning. Gentamicin sulfate solution 0.3% was more effective than chloramphenicol 0.5%, sulphamethizole 4%, bacitracin 1%-neomycin 0.5%, and ristocetin 0.5%-polymyxin B 0.25% solutions, and oxytetracycline 3%-polymyxin B 0.1% ointment. Whitney and coworkers130 demonstrated in 1972 a significant reduction in the incidence of Staphylococcus species from the eyelids preoperatively with the use of antibiotics, especially gentamicin 1.0% solution. (Note that this gentamicin concentration was more than three times that of the currently commercially available concentration.) Gentamicin was superior to chloramphenicol 0.5% solution, polymyxin B-neomycin-gramicidin solution, and polymyxin B-neomycin-bacitracin ointment in this study. Six doses given hourly of topical 0.3% ciprofloxacin has been shown to significantly lower counts of coagulase-negative staphylococci and total bacteria preoperatively in normal eyes.131 Osher and coworkers55 showed that one drop of either 0.1% trimethoprim/polymyxin B sulfate (Polytrim), 10,000 U/ml, or 0.3% tobramycin given 4 times daily for 2 days and 90 minutes and 60 minutes before cataract extraction eradicated all organisms identified in baseline conjunctival cultures except S epidermidis. Many surgeons apply antibiotics topically preoperatively for endophthalmitis prophylaxis; however, the data supporting this are somewhat limited. Dunnington and Locatcher-Khorazo113 performed an early retrospective analysis of postoperative endophthalmitis in cataract surgery and found no cases of endophthalmitis in 663 patients treated with either topical penicillin or sodium sulfathiazole ointments preoperatively. A nonconcurrent historical control group treated with silver-protein solution had an endophthalmitis incidence of 1.8%. Another early study described 7095 cataract patients who received various topical antibiotics preoperatively.114 An incidence of endophthalmitis of 0.08% in treated patients was thought to be an improvement over historical controls. In a nonrandomized trial spanning 21 years, patients given topically applied antibiotics preoperatively for cataract surgery had a lower incidence of endophthalmitis than patients not given antibiotics.132,133 Newer, more powerful antibiotics—specifically the fluoroquinolones—have enjoyed increasing popularity for prophylactic topical uses in ophthalmology at the possible expense of increasing resistance to this class of antimicrobial agents. For this reason, their prophylactic use is still controversial and should be considered carefully in formulating a prophylactic regimen. The effect of subconjunctival antibiotics on endophthalmitis has been studied in large groups. A study of 54,000 patients in the Indian subcontinent showed no decrease in endophthalmitis in those patients given antibiotics subconjunctivally, compared with those who were not.134 A subsequent study on 23,900 patients showed a reduction in endophthalmitis when antibiotics were given subconjunctivally and topically, compared with topically only.135 The dangers of administering antibiotics subconjunctivally include inadvertent penetration of the globe and entrance of antibiotic through sutureless corneoscleral incisions. Indeed, episodes of sterile endophthalmitis may be secondary to toxic levels of antibiotics entering the eye by direct intracameral instillation at the end of surgery or through sutureless wounds after postoperative subconjunctival injection.136–140 Various investigators have advocated the addition of prophylactic antibiotics in irrigating solutions used during cataract surgery to overcome barriers to intraocular drug penetration by extraocular drug administration routes, although this remains controversial.141–144 A randomized, prospective trial in 97 patients showed no significant reduction in the rate of positive bacterial cultures from anterior chamber isolates when gentamicin, 8 μg/ml, was added to the irrigating solution during phacoemulsification.145 A double-blind placebo-controlled randomized trial comparing the effect of vancomycin, 20 μg/ml, and gentamicin, 8 μg/ml, to placebo in the irrigating fluid during phacoemulsification in 120 eyes showed a nonsignificant reduction in positive anterior-chamber cultures in the antibiotic group.146 Gritz and coworkers147 studied the effect of suspending various bacteria in solutions of balanced salt solution (BSS); BSS plus vancomycin, 20 μg/ml; BSS plus gentamicin, 8 μg/ml; and BSS plus vancomycin and gentamicin for 30, 60, and 120 minutes to simulate antibiotic addition to irrigating solutions. Only 2 of 12 organisms, P aeruginosa and Moraxella nonliquifaciens, were affected by antibiotic exposure in vitro for these periods. It appears that the addition of antibiotics to irrigating solutions during cataract surgery is of questionable benefit; there are no large clinical trials proving its efficacy in decreasing the incidence of postoperative endophthalmitis. Endophthalmitis has been reported even when antibiotics were present in the irrigating solution.148 The addition of vancomycin may actually be deleterious in the long term, considering emerging VRE and reduced susceptibility to vancomycin in S aureus (VISA).31,32,149 Because antibiotic susceptibilities may change with increasing use of certain antibiotics such as the aminoglycosides (including gentamicin), the relative effectiveness of the antibiotics noted in many of the older studies may differ from results found in more recent studies. Increased resistance to aminoglycosides has been noted among Staphylococcus isolates causing postoperative endophthalmitis in recent years.103 The incidence of infection after intraocular surgery has shown a steady decline with advances in surgical technique, equipment, and the use of prophylactic antibiotics.46 The incidence of endophthalmitis after ophthalmic surgery was more than 1% in the first half of this century.45 A study of 36,000 cataract patients in 1974 estimated the incidence of endophthalmitis at 0.086%, and a more recent study of 23,625 extracapsular cataract extractions in 1991 reported a similar incidence of 0.072%.132,150 Although there are no definitive data, the widespread use of prophylactic topical, subconjunctival, and intracameral (in irrigating solutions) antibiotics may have affected endophthalmitis rates. ANTISEPTICS Antiseptics applied to skin for prophylaxis have been standard for some time and have an advantage over antibiotics in the relative immediacy of their bactericidal action and the absence of promotion of bacterial resistance to antibiotics. In 1951, testing of various preparations showed 3% hexachlorophene, followed by a sterile saline rinse and 3.5% iodine and 70% alcohol, to be the most effective.151 Hexachlorophene is no longer in wide usage because of its neurotoxicity and poor activity against gram-negative bacteria.152 Chlorhexidine is a disinfectant with broad efficacy but it can cause severe corneal damage and should be avoided in ophthalmic skin preparations.152,153 Iodine is an excellent antiseptic; when combined with povidone (polyvinyl pyrrolidone), it is a safe periocular skin and conjunctival disinfectant.154 Preoperatively, 5% povidone-iodine placed in the conjunctival cul-de-sac has been shown to be safe and in one study reduced conjunctival bacterial colonies by 91% and species by 50%.155,156 There is also evidence that there is a synergistic effect on postoperative infection rates when it is used in conjunction with topical antibiotics.157,158 Postoperatively applied povidone-iodine prevents an increase in the number of colony-forming units and lowers the conjunctival species counts at 24 hours after surgery.159 It also lowers colony-forming units and slows the increase in species counts at 1 week after surgery.160 In a nonrandomized trial, povidoneiodine lowered the incidence of postoperative endophthalmitis when compared with silver-protein solution.161 |
PREVENTION OF POSTTRAUMATIC INFECTIONS |
PREVENTION OF POSTTRAUMATIC ENDOPHTHALMITIS Even if small and self-sealing, any penetrating ocular injury has the potential to introduce microorganisms. Although not common, traumatic endophthalmitis can be devastating; therefore, antibiotic prophylaxis is justified. There are no randomized trials clearly demonstrating their effectiveness, however, and a retrospective analysis of endophthalmitis in penetrating trauma showed no benefit to antibiotic use.162 Endophthalmitis after trauma is associated with a worse visual outcome than after surgery.163 It is important to emphasize that the diagnosis of traumatic endophthalmitis may be delayed by attributing signs of inflammation solely to trauma.164,165 An increased relative risk of infection has been demonstrated in eyes with disruption of the crystalline lens, delayed primary repair, intraocular foreign body, and a plant- or soil-related injury in a rural environment.162,164,166,167 The organisms associated with traumatic endophthalmitis reflect the source of injury and are often different from the causes of postoperative endophthalmitis. The most common agent is S epidermidis, followed by Bacillus species, Streptococcus species, S aureus, gram-negative bacilli, and fungi.164,168–171 Posttraumatic endophthalmitis is more likely to be associated with mixed microbial infections than postoperative endophthalmitis.163,169 After obtaining cultures, empiric broad-spectrum antibiotics are warranted, with particular attention to Bacillus species (especially B cereus)—one of the most virulent organisms encountered in this setting.165,172–176 Bacillus species may be found in postoperative or endogenous endophthalmitis but are more common after penetrating eye injury, especially when associated with soil contamination or endogenous infections in intravenous drug users.171–174,177–180 Topical, subconjunctival, and intravenous antibiotics are routinely used prophylactically; however, their relative benefit has not been established in a controlled trials. Some authors recommend intraocular (intravitreal and anterior chamber) antibiotics but injection of antibiotics remains controversial because of the risks associated with poor visualization often encountered in the trauma setting.110,164,173,181,182 A prospective study of 30 ruptured globes found that the use of intravenous antibiotics significantly reduced the occurrence of organisms in the anterior chamber at surgery.183 Generally, broad-spectrum antibacterial coverage against gram-positive cocci and gram-negative bacilli is warranted with the addition of Bacillus species coverage if there is an intraocular foreign body or soil contamination. In the latter instance, use of vancomycin or clindamycin in addition to gentamicin may be appropriate, despite the general discouragement in using vancomycin for prophylactic indications. This would be true considering the often devastatingly poor visual outcome in cases of Bacillus species endophthalmitis. A review of 36 cases of Bacillus species infection found universal susceptibility to aminoglycosides and vancomycin, variable susceptibility to clindamycin, and relative resistance to cephalosporins.184 PREVENTION OF OCULAR TRAUMA-ASSOCIATED TETANUS Clostridium tetani is an anaerobic spore-forming bacterium and is the cause of tetanus. The incidence of tetanus in the United States has fallen markedly from 560 cases in 1947 (the year national reporting was initiated) to 48 cases in 1987.185 This decrease is attributed to better wound cleaning and hygiene and the initiation of prophylaxis with tetanus toxoid. Tetanus infection increases with age, with most cases occurring in those older than 50.185 The recommendation for primary immunization in children, ages 6 weeks through 6 years (up to the 7th birthday), is intramuscular administration of tetanus toxoid combined with diphtheria and pertussis on four occasions.186 The first three doses are at 4- to 8-week intervals, and the fourth dose is given about 6 to 12 months after the third dose. Boosters are given at age 4 to 6 years and then every 10 years after the last dose. Primary immunization at age 7 or older is without pertussis (Td) because the severity of pertussis decreases with age, and adverse neurologic reactions to pertussis have been reported in adults.186 Three doses of Td are given: the second is 4 to 8 weeks after the first, and the third dose is 6 to 12 months after the second. Booster is given every 10 years after the last dose. Patients with unknown or uncertain vaccination histories are assumed to have had no previous tetanus toxoid.186 If those patients present with a clean minor wound, they are given primary immunization. If the wound is contaminated (dirt, feces, soil, or saliva) or major (a puncture wound or a wound from missiles, crushing, burn, or frostbite), primary vaccination is given along with 250 U of human tetanus immune globulin intramuscularly. If both tetanus toxoid and tetanus immune globulin are given concurrently, separate syringes and sites should be used. If the primary vaccination has been completed, a booster of tetanus toxoid is given when more than 5 years has elapsed since the last dose and the wound is severe or dirty. If the wound is clean and minor, a booster is given if 10 years have elapsed since the last dose. More frequent boosters are not needed and can accentuate side effects. The Immunization Practices Advisory Committee does not specifically address ocular injuries such as corneal abrasions. A study in an animal model showed that unimmunized mice developed clinical tetanus after injection of C tetani organisms or toxin into the anterior chamber.187 These mice, however, did not develop tetanus with topical inoculation after corneal debridement or stromal scarification. Immunized mice did not develop clinical tetanus after intracameral injection or topical inoculation. These data suggest prophylaxis against tetanus is indicated after perforating injuries but not after nonperforating injuries such as corneal abrasions. Ostler188 has recommended treating corneal injuries caused by metallic foreign bodies and having minimal tissue destruction as “minor” and corneal injuries caused by or associated with organic matter or dirt or with extensive necrosis as “major.” CORNEAL ABRASION Corneal abrasions are common and found in about 10% of eye emergency visits.189 Traditionally, noninfected noncontact lens-related corneal abrasions are treated with a topical antibiotic for infection prophylaxis—sometimes with a topical cycloplegic for patient comfort and a pressure patch. The benefit of pressure patching is questioned.190–195 To the best of our knowledge, there are no randomized double-blind placebo-controlled trials looking at the advantage of prophylactic antibiotics for noninfected corneal abrasions. Because the incidence of microbial keratitis in this setting is so low, it is unlikely that a study such as this will be performed. Topical fluoroquinolones enjoy wide usage by ophthalmologists, and ofloxacin has been shown to be as effective as tobramycin for the treatment of external ocular infection—as effective as fortified cefazolin and tobramycin for bacterial keratitis and as effective as fortified gentamicin and cefuroxime in microbial keratitis.190,196,197 Fluoroquinolones are probably the most common agent used for prophylaxis in corneal abrasions because of their broad-spectrum, low toxicity, and low resistance levels in commonly acquired organisms. Prolonged and low-frequency dosage, however, should be avoided to discourage emergence of resistant organisms in the presence of subinhibitory antibiotic levels on the ocular surface.36 The annual incidence of ulcerative keratitis is estimated at 0.13% to 0.21% for extended-wear soft contact lens wearers and 0.02% to 0.04% for daily-wear soft contact lens wearers; Pseudomonas is the most common cause in these lens wearers.198–202 Patching for corneal abrasions in contact lens wearers is not recommended because of potential incubation of infecting organisms and promotion of subsequent infectious keratitis.203 SUTURE REMOVAL AND CUTTING Although it is common practice to administer prophylactic antibiotics before and after cutting or removing sutures placed in the globe, a search of the literature does not reveal data to support this practice. There have been case reports of endophthalmitis after trabeculectomy with releasable sutures, and it has been suggested that minimizing exposure of suture material reduces the risk of infection in this setting.204–206 Suture abscesses can occur after penetration keratoplasty, especially because these patients often administer topical steroids for extended periods.207 Despite the absence of supportive data, prophylaxis at the time of corneal, corneoscleral, or scleral suture cutting or removal seems prudent, with appropriate considerations given to likely infecting organisms and appropriate antibiotic choices. In addition, care should be taken to avoid pulling exposed suture material through the wound. |
BULLOUS KERATOPATHY |
Bullous keratopathy (BK) remains one of the most common indications for penetrating keratoplasty (PKP) in the United States; for about the past two decades, about 25% of PKPs were the result of of pseudophakic BK.208–211 BK is seen in about 6% to 9% of patients who develop microbial keratitis.212–214 The incidence of microbial keratitis in BK patients is less well known, although estimates are between 1.8% and 4.7%.215,216 Luchs and coworkers216 found that steroids, bandage contact lens use, bullae, and antibiotic use significantly increased the risk of ulcerative keratitis in BK in a retrospective review. As the authors note, the paradoxical increase in episodes of infection in patients who used prophylactic antibiotics may be because of bias selection; physicians may be more likely to give antibiotics to eyes that have recurrent epithelial defects and are thus at increased risk of developing infection. The indications for prophylactic antibiotics in BK are inconclusive, although it is clear that antibiotics do not prevent the development of ulcerative keratitis in many cases. |
TISSUE TRANSPLANTATION MATERIAL |
The risk of endophthalmitis after PKP is estimated between 0.1% and 0.77%.150,217–225 The Eye Bank Association of America recommends that routine donor corneoscleral
rim cultures be performed at the time of PKP.226 Positive cultures occur at an estimated rate of 4% to 39%.217,218,220,222,223,227–230 Common organisms isolated from culture include S epidermidis, Propionobacterium acnes, S aureus, and Streptococcus viridans.217,218,220–224,227–230 Corneal storage media include gentamicin, in an effort to reduce the incidence of endophthalmitis.231 There have been significant reports of gentamicin resistance among corneal rim culture isolates and in endophthalmitis after PKP, however.217,218,220–223,227–229 Some advocate the addition of more broad-spectrum coverage, especially for gram-positive organisms, and the stability and safety of vancomycin with corneal endothelial cells has been established.232,233 Hwang and associates234 investigated the in vitro effect of 11 different antibiotics under simulated corneal storage conditions and found streptomycin to be the most effective against S aureus and S epidermidis. Further studies showed an improvement in activity against gentamicin-sensitive and gentamicin-resistant bacterial isolates when streptomycin, 200 μl, was added to Optisol (Chiron Vision, Irvine, CA) containing gentamicin, 100 μl.235 Because storage-media antibiotics require actively growing bacteria to exert their cidal effects, some authors advocate warming donor corneas to room temperature preoperatively to achieve this effect. Several hours of warming may be necessary to achieve bacterial growth and antibiotic killing after refrigeration of donor material. Other antibacterial storage media have been described for storage of human sclera, including glycerol, ethanol, and Zephiran (Sanofi, New York, NY).236 Because of the ready availability of human donor tissue on short notice, however, the Eye Bank Association of America abandoned glycerin as storage media for emergency situations in 1992 and from 1993 onward, only Optisol-preserved tissue has been available to corneal surgeons.237 A review of 1876 cases of PKP in 1983 showed positive corneal rim cultures in 230 (12%) and endophthalmitis in 4 (0.2%).218 Cultures were positive in 3 of the 4 cases of endophthalmitis. The incidence of endophthalmitis in eyes with positive cultures (1.3%) was 22 times higher than in eyes with negative cultures (0.06%); therefore, the authors recommended culturing of all scleral rims and coverage of all cultured organisms with topical antibiotics. Donor corneas were stored in McCarey-Kaufman media with gentamicin sulfate, 100 μg/ml, and patients received topical antibiotics before and after surgery (type and duration of treatment were not specified). The causative organisms were S aureus, P aeruginosa, Enterococcus fecal, and S pneumoniae. Three of these organisms were resistant to gentamicin. A retrospective review of 1078 penetrating keratoplasties performed at the Mayo Clinic between 1981 and 1995 showed positive corneal rim cultures in 209 cases (19.4%) and endophthalmitis in 3 cases (0.28%).220 Rim cultures were negative in all cases of endophthalmitis. Causative organisms in endophthalmitis were S aureus and S pneumoniae. Patients received either chloramphenicol (before 1989) or neomycin-polymixin B-gramicidin (after 1989) drops the night before surgery and after surgery until the cornea re-epithelialized. Subconjunctival gentamicin sulfate and methylprednisolone injections were given at the end of surgery. The authors calculated a positive predictive value for endophthalmitis of 0.0, based on corneal donor rim cultures. Moreover, they calculated low positive predictive values (between 0.012 and 0.024) for endophthalmitis in three other large studies after PKP.218,221,223 Finally, they concluded that preoperative corneal rim cultures are of little or no benefit and add unnecessary expense. |
CONJUNCTIVITIS IN THE NEWBORN |
Conjunctivitis is the most common ocular disorder of newborns.238 Infection from Neisseria gonorrhoeae is potentially the most severe because it can be blinding. Other rarely
sight-threatening causes include Chlamydia trachomatis, S epidermidis, S viridans, S aureus, Haemophilus
influenzae, herpes simplex, P aeruginosa, and chemical (silver nitrate).239 Credé introduced the use of 2.5% silver nitrate in newborns as prophylaxis against gonorrheal conjunctivitis in 1881.1 With this intervention, there was a marked decrease in incidence, although N gonorrhoeae still remained a threat.240 N gonorrhoeae is implicated in 0.3% of neonatal conjunctivitis.241 Silver nitrate causes chemical conjunctivitis in many newborns and is presumably not effective against chlamydia;242–245 therefore, tetracycline and erythromycin ointments are used more commonly in developed countries. It is important to note that topically applied ocular medication has little or no effect on chlamydial nasopharyngitis or pneumonia, so topical prophylaxis against conjunctivitis may delay the diagnosis of pneumonia and thereby have a deleterious effect on long-term pulmonary function. A randomized trial comparing 1% silver nitrate drops and 1% tetracycline ointment for prophylaxis against gonococcal and chlamydial conjunctivitis in 2732 newborns was performed in Kenya in the mid-1980s.246 The incidence of gonococcal and chlamydial conjunctivitis after silver nitrate prophylaxis was 0.4% and 0.7%, respectively, whereas after tetracycline it was 0.1% and 0.5%. These rates were a significant reduction from those in historical controls of 2.7% and 6.2%. Although the rates with tetracycline were lower than with silver nitrate, they did not reach statistical significance. Because of a broader spectrum of activity, lower cost, and no reported toxicity to the cornea and conjunctiva, povidone-iodine has been used as prophylaxis against neonatal conjunctivitis.247 A randomized study of 100 newborns given 2.5% povidone-iodine solution in one eye and either 1% silver nitrate solution or 0.5% erythromycin ointment in the other eye was performed in 1994.248 Povidone-iodine had the most significant reduction in the number of colony-forming units and number of species. Silver nitrate had significantly more ocular toxicity than povidone-iodine or erythromycin. Isenberg and associates249 compared 2.5% povidone-iodine solution, 1% silver nitrate solution, and 0.5% erythromycin ointment in a randomized trial involving 3117 newborns in Kenya. There was a significant reduction in the incidence of infectious conjunctivitis in the povidone-iodine group, compared with silver nitrate or erythromycin. Povidone-iodine was significantly more effective against C trachomatis than either silver nitrate or erythromycin; there were no differences in incidence of N gonorrhoeae or S aureus in the three treatment groups. Povidone-iodine costs less, is less toxic, and appears to be more effective than silver nitrate or erythromycin. A literature search reveals no existing trials comparing povidone-iodine with tetracycline ointment. |
HERPETIC KERATITIS: ORAL ACYCLOVIR FOR RECURRENT KERATITIS |
Herpes simplex virus (HSV) keratitis is one of the most common indications
for PKP in the United States.250 Oral acyclovir has been shown to reduce recurrences of both labial and
genital infections but its effect on the recurrence of ocular herpes
is less well known.251,252 Studies evaluating the effect of acyclovir on acute keratitis and on postoperative
penetrating keratoplasty have shown some beneficial effects.253–257 Recurrence rates of herpetic (epithelial) keratitis are estimated to be
one episode every 6 months to 3 years but the issue of long-term oral
acyclovir as prophylaxis against recurrences remains controversial, with
little supporting data.258,259 A randomized placebo-controlled trial of short-term acyclovir (3 weeks) for
the prevention of stromal keratitis or iritis in patients with HSV
epithelial keratitis showed no benefit of acyclovir.260 Patients in the acyclovir group had fewer recurrences of epithelial keratitis
while taking acyclovir but more recurrences during the remainder
of the study period (12 months). A nonrandomized nonstandardized study of the effect of long-term (mean, 27 months) prophylactic acyclovir on 13 patients with a history of frequently recurring herpes simplex keratitis showed a reduction in the rate and duration of recurrences.259 A definitive answer to this question cannot be found, however, until a randomized placebo-controlled study is performed. This is underway as the Acyclovir Prevention Trial portion of the Herpetic Eye Disease Study.261 |
CYTOMEGALOVIRUS PROPHYLAXIS |
The risk of developing CMV retinitis in acquired immunodeficiency syndrome (AIDS) is estimated to be between 15% and 40%.262–267 Standard induction therapy for established retinitis is with intravenous ganciclovir or foscarnet sodium, and most patients relapse within 2 months.268–273 A prospective, randomized, double-blind placebo-controlled study was performed on the effectiveness of ganciclovir for prophylaxis against the development of CMV disease in CMV-infected patients with AIDS.274 Median CD4 count was 22/mm3 and CMV-infection was confirmed by serologic or urine testing. In a 2-to-1 ratio, patients received either oral ganciclovir, 100 mg 3 times daily, or placebo. After a median of 367 days, the study was discontinued because there was a 49% reduction in CMV disease in the ganciclovir group (p < 0.001). This study shows a significant reduction in the risk of CMV in patients with AIDS who take oral ganciclovir as prophylaxis. |
HUMAN IMMUNODEFICIENCY VIRUS CHEMOPROPHYLAXIS AFTER MUCOUS MEMBRANE EXPOSURE |
The risk of transmission of HIV after a percutaneous injury is estimated at 0.2% to 0.3%; the risk after mucocutaneous injury (including conjunctiva) is probably less, although estimates are somewhat unreliable.275–278 Although rare, transmission of HIV may occur after exposure of mucous membrane (mouth) to contaminated blood.279 A case-controlled study of patients who seroconverted after percutaneous exposure to HIV identified several risk factors associated with HIV transmission: deep injury, injury with a device visibly contaminated with the source patient's blood, procedures involving a needle placed in the source patient's vein or artery, and terminal illness in the source patient.280 Seroconverters were less likely to have taken zidovudine than controls. Although all of the case and control patients in this study had exposure by needle sticks or other sharp objects, these risk factors for HIV seroconversion probably also apply after conjunctival exposure. Recommendations for chemoprophylaxis after mucous membrane exposure with blood or fluid containing blood is zidovudine (AZT or ZDV), 200 mg 2 times daily; lamivudine (3TC), 150 mg 2 times daily; and possibly indinavir (IDV), 800 mg 3 times daily if exposure is with a large volume of blood or the blood has high HIV titer.281 This regimen is continued for 4 weeks. |