Responsiveness of Experimental Surgical-Wound Infections

ANTIMICROBIAL AGENTS AND CHZMOTHERAPY, JUlY 1976, P. 45-51 Copyright C 1976 American Society for Microbiology Vol. 10, No. 1 Printed in U.S.A. Responsiveness of Experimental Surgical-Wound Infections to Topical Chemotherapy R. J. McRIPLEY* AND R. R. WHITNEY Department of Microbiology, The Squibb Institute for Medical Research, Princeton, New Jersey 08540 Received for publication 9 February 1976 Topical agents freshly formulated in a cream base vehicle as well as commercial topical preparations were used to evaluate in mice the responsiveness of experimental surgical wounds infected with Staphylococcus aureus or Pseudomonas aeruginosa to chemotherapy. The responsiveness of the infections to therapy or the efficacy of a topical agent was assessed primarily by means of wound counts of the infecting organism before and after the employment of an immediate (prophylactic) or delayed (therapeutic) treatment regimen. From tests of several concentrations of an agent formulated in the vehicle, a median effective dose could be determined. In the case of the lethal P. aeruginosa infection, a median protective dose could be determined. Both infections were found to be quite susceptible to treatment with those topical agents that demonstrated good activity in vitro against the test organisms. The results of the investigation indicated that the model infections were suitable for the screening of potential topical agents in vivo. In the preceding paper experimental surgical-wound infections in mice were described and characterized (6). In addition, a rapid method for quantitating bacteria in the wound infections was presented. In this paper the response of the surgical-wound infections to chemotherapy with various topical antimicrobial agents is explored. Freshly formulated antibacterial agents, commonly administered topically, as well as commercial topical antimicrobial products were employed in the study. These topical formulations were used not only for evaluating the responsiveness of the experimental infections to chemotherapy but also for demonstrating how, using standardized methodology, the model infections can be employed to assess various measures oftopical efficacy. MATERIALS AND METHODS Antibacterial agents. The antibiotics and chemicals used in this study were gentamicin sulfate, 634,ug/mg (Schering Corp.); polymyxin B sulfate, 7,956 U/mg (Pfizer Laboratories); neomycin sulfate, 654,ug/mg (E. R. Squibb & Sons); zinc bacitracin, 47.7 U/ mg (Commercial Solvents); gramicidin, 943 jag/mg (E. R. Squibb & Sons); nitrofurazone (Eaton Laboratories); and triclobisonium chloride (Roche Laboratories). The agents were supplied as powders or crystals and, for testing in vivo, were formulated at the desired concentration on a weight basis in an emulsion-type cream base vehicle containing promulgen, protopet, sorbitol solution, silicone antifoam, distilled water, and methyl and ethyl parabens. Commercial products. These products, with the 45 active ingredients listed, were purchased on the open market: Neosporin ointment (Burroughs Wellcome Co.), consisting of (per gram) 5,000 U of polymyxin B sulfate, 400 U of zinc bacitracin, and neomycin sulfate equivalent to 3.5 mg ofneomycin base; Mycolog cream (E. R. Squibb & Sons), consisting of (per gram) 100,000 U of nystatin, 0.25 mg ofgramicidin, and neomycin sulfate equivalent to 2.5 mg of neomycin base, including 0.1% triamcinolone acetonide; Polysporin ointment (Burroughs Wellcome Co.), consisting of (per gram) 10,000 U of polymyxin B sulfate and 500 U of zinc bacitracin; Furacin topical cream (Eaton Laboratories), consisting of 0.2% nitrofurazone; Garamycin cream (Schering Corp.), consisting of 0.1% gentamicin sulfate; and Triburon ointment (Roche Laboratories), consisting of 0.1% triclobisonium chloride. Evaluation of antibacterial activity in vitro. Minimal inhibitory concentrations (MIC) were determined by the conventional twofold dilution method, using antibiotic assay broth (BBL) as previously described (3). Aliquots from all tubes showing no visible growth were plated on antibiotic assay seed agar (BBL) to determine the minimal bactericidal concentration (MBC). Antibacterial agents were dissolved in 0.05 M phosphate buffer, ph 7.0, prior to being evaluated in vitro. Evaluation of efficacy in vivo. Details of the experimental Staphylococcus aureus and Pseudomonas aeruginosa surgical-wound infections were presented in the preceding paper (6). The efficacy of test formulations in the experimental infections was assessed by two methods: (i) inhibition or reduction of viable wound counts of both test organisms, and (ii) in the case of the P. aeruginosa infection, protection from the lethal effect of the infection. When quanti-

46 McRIPLEY AND WHITNEY tation of viable organisms was utilized, two treatment regimens were employed: (i) immediate (prophylactic) treatment and (ii) delayed (therapeutic) treatment. With the immediate-treatment regimen, which was initiated prior to the actual establishment of wound infections, each wound was treated topically with 0.4 g of formulation 15 min after the insertion of sutures and then 6 h later. The delayedtreatment regimen, which commenced after wound infections were established, consisted of two treatments administered 24 and 30 h after suture insertion. With both regimens, wounds were cultured 18 h after the final treatment using the surface rinse technique (6), and counts of viable infecting organisms were made by conventional methods. Ten mice were used for evaluating the efficacy of each formulation. When the immediate-treatment regimen was used, the efficacy of a formulation was based on its ability to inhibit the establishment of wound infections as reflected by the populations of test organisms in wounds after treatment relative to those obtained in placebo-treated wounds. The difference between the means of the two populations was assumed to represent the inhibition produced by an active agent in the formulation. When the delayed-treatment regimen was used, a formulation's efficacy was assessed by the reduction in the bacterial wound counts produced by treatment relative to pretreatment counts of the test organism. In evaluating the protective efficacy of test formulations against the lethal P. aeruginosa infection, all mice, 10/group, were treated with cream base immediately after infection. After 6 h, treatment with test formulations commenced. Each mouse was treated topically with 0.4 g of formulation. Treatment continued twice daily for the next 2 days. Throughout the course of therapy, mice were closely observed to ascertain whether the topically applied formulations were being ingested. Mice were held for observation a total of 2 weeks, with deaths being recorded daily. The median effective dose (ED,e) or the median protective dose (PDO) of an antibacterial agent was determined by the method of Reed and Muench (8) and was based on a minimum of three dosage levels at fourfold increments. Detection of drug activity in serum and wound washings. At 18 h after the termination of therapy, blood and wound washings were obtained from noninfected mice treated with each test drug at the highest concentration used in the study. The washings and the collected sera were passed through 0.45-,um membrane filters (Millipore Corp.) prior to being tested for drug activity. Aliquots of washings and sera were incubated at 37 C in the presence of known concentrations of P. aeruginosa or S. aureus (103 to 105 colony-forming units/ml). At various intervals during the incubation period, samples were removed from each mixture, diluted appropriately, and cultured quantitatively on brain heart infusion agar. The mean concentration of viable organisms determined for each group of test mixtures was compared with those concentrations found with control mixtures that contained sterile washings or sera from placebo-treated or untreated mice. In addition, ANTIMICROB. AGZNTS CHZMOTHZR. aliquots of wound washings and bacterial inocula containing 50 to 300 viable organisms were placed on brain heart infusion agar plates and spread over the surface of the agar. The plates were incubated at 37 C for 48 h. The number of colonies formed after incubation was compared with control counts obtained when washings were omitted. RESULTS The activity in vitro of several antibacterial agents, commonly utilized topically, against the two test organisms used in the surgicalwound infections is shown in Table 1. The results indicate that the agents tested varied considerably in their activity against the test organisms. Against S. aureus, triclobisonium and gentamicin were the most active agents tested, exhibiting MIC values of <1.0,ug/ml. Nitrofurazone and neomycin also demonstrated good activity, whereas gramicidin and polymyxin B were less active. Against P. aeruginosa, gentamicin and polymyxin B were the most active agents. Neomycin and triclobisonium were also active, whereas the other agents were essentially inactive. With the exception of the activity of neomycin and gentamicin against P. aeruginosa, MBC values for the various agents were not markedly different from MIC values. Several commercial topical preparations were evaluated in the experimental infections to determine the responsiveness of the infections to treatment with products specifically marketed for use against local infections. The results (Table 2) indicate that most of the products were efficacious against the S. aureus infection. All the products except the nitrofurazone formulation were extremely effective in inhibiting the growth of the test organism in the wound when the immediate-treatment schedule was followed; the nitrofurazone prod- TA.LE 1. Activity in vitro of selected antibacterial agents used in commercial topical products Test organism Antibacterial agent mic gl Ml)C S. aureus Gentamicin sulfate 0.78 1.6 (SC 2406) Gramicidin 18.7 25 Neomycin sulfate 6.25 12.5 Nitrofurazone 9.4 12.5 Polymyxin B sulfate 37.5 75 Triclobisonium chlorid 0.6 1.6 Zinc bacitracin 6.25 12.5 P. aeruginosa Gentamicin sulfate 0.6 3.1 (SC 8822) Gramicidin >100 >100 Neomycin sulfate 9.4 50 Nitrofurazone >100 >100 Polymyxin B sulfate 1.6 3.1 Triclobisonium chloride 25 75 Zinc bacitracin >100 >100

VOL. 10, 1976 TOPICAL CHEMOTHERAPY 47 TABLz 2. Activity ofselected commercial topical products against experimental surgical-wound infections in mice Mean inhibitiona Mean reduc- Experimental infection Formulation (log,o) tionb(loglo) S. aureus Gentamicin cream 5.02 ± 0.24e 3.40 ± 0.25 Neomycin-gramicidin-nystatin- 5.82 + 0.19 3.85 ± 0.43 triamcinolone acetonide cream Nitrofurazone cream 2.85 + 0.23 0 Polymyxin B-bacitracin-neomycin 5.54 ± 0.18 3.92 + 0.40 ointment Triclobisonium chloride ointment 4.99 ± 0.59 3.77d P. aeruginosa Gentamicin cream 5.33 + 0.14 2.29 ± 0.33 Neomycin-gramicidin-nystatin- 4.42 + 0.32 0.43 ± 0.25 triamcinolone acetonide cream Nitrofurazone cream 0.36 ± 0.09 0 Polymyxin B-bacitracin-neomycin 4.02 ± 0.21 0.69 ± 0.03 ointment Triclobisonium chloride ointment 2.59 ± 0.28 0.49 ± 0.14 a Inhibition relative to the mean concentration of viable test organisms in placebo-treated wounds; immediate treatment. b Reduction relative to the mean wound concentration of viable test organisms prior to treatment; delayed treatment. c Mean ± standard error of a minimum of three experiments. d Tested only once; product is no longer commercially available. uct was less effective than the other products. In the more stringent evaluation that occurred when treatment was delayed, all the products except the nitrofurazone preparation produced a significant reduction in the S. aureus wound count. The products were more diversified in their activity against the experimental P. aeruginosa infection. In the evaluation of prophylactic activity, i.e., with the immediatetreatment schedule, gentamicin and the two combination products were quite effective, the triclobisonium ointment was moderately active, and the nitrofurazone cream was essentially inactive. However, under the rigorous assessment of bactericidal efficacy that occurred when treatment was delayed, only the gentamicin formulation was efficacious. To evaluate further the effectiveness of the experimental infections in assessing the efficacy of topical antibacterial agents, several of the agents used in commercial products were formulated at various concentrations near the "use" level in a compatible cream base vehicle and were tested in the model infections. The results of this abbreviated dose response study are presented in Table 3. In general, the two infections were quite responsive to therapy with the various series of formulations and demonstrated approximately 2 log1o differences in bacterial wound counts when treated with two or three 10-fold decrements of the agent being evaluated. Neomycin formulations were the notable exception to this observation, exhibiting modest differences in wound counts in the S. aureus infection when applied prophylactically and in the P. aeruginosa infection when applied therapeutically. As was expected, with delayed treatment, larger concentrations of the various agents were required to produce significant responses in the infections than were required with the immediate-treatment regimen. For a definitive comparison of the efficacy of different antibacterial agents in the experimental infections, a method of assessing activity based on the EDO was employed. At least three concentrations of an agent, formulated at fourto fivefold increments in the cream base vehicle, were tested. The EDO was defined as the concentration that produced a 3 log10 inhibition or reduction in the wound count of the infecting organism in 50% of mice. The results of a typical experiment are presented in Table 4. Against the S. aureus infection, gentamicin and neomycin demonstrated equivalent activity when the immediate-treatment schedule was followed; however, gentamicin was more efficacious than neomycin when delayed treatment was employed. Bacitracin was not as efficacious as either of the other two antibiotics. Against the P. aeruginosa infections, polymyxin B was the most efficacious agent, demonstrating the same ED50 with both the immediate- and delayed-treatment schedules. Gentamicin was more active than neomycin but less active than polymyxin B. In using inhibition and reduction of bacterial wound counts as criteria of formulation effi-

48 McRIPLEY AND WHITNEY TABrz 3. ANTIMICROB. AGENTS CHEMOTHER. Effect of concentration on the activity of selected antibacterial agents in surgical-wound model infections in mice Experimental infection Test drug Concn ( Mean inhibition" Mean (log,0) (log,,) reductione S. aureus Gentamicin sulfate 0.1 5.16 ± 0.36c 2.88 ± 0.22 0.01 2.40 ± 0.70 0.36 ± 0.15 0.001 0.96 ± 0.05 Neomycin sulfate 0.5 4.90 ± 0.05 3.30 ± 0.37 0.05 3.84 ± 0.20 1.59 ± 0.25 0.005 2.37 ± 0.71 Zinc bacitracin 1.0 0.43 ± 0.30 0.5 4.49 ± 0.27 0.05 1.80 ± 0.32 P. aeruginosa Gentamicin sulfate 0.5 2.80 ± 0.45 0.1 4.92 ± 0.16 0.05 0.49 ± 0.24 0.01 2.23 ± 0.42 0.001 0.42 ± 0.15 Neomycin sulfate 2.0 2.87 ± 0.11 1.0 4.50 ± 0.22 0.2 1.25 ± 0.28 0.1 1.47 ± 0.21 Polymyxin B sulfate 0.05 5.20 ± 0.63 3.55 ± 0.17 0.005 1.40 ± 0.20 0.43 ± 0.18 a Inhibition relative to the mean concentration of viable test organisms in placebo-treated wounds; immediate treatment. b Reduction relative to the mean wound concentration of viable test organisms prior to treatment; delayed treatnent. I Mean ± standard error of two to four experiments. TABLz 4. ED,0 of sekected antibacterial agents against surgical-wound infections based on the inhibition or reduction in counts of viable infecting organisms ED.0 (%) S. aureus P. aeruginosa infection infection Test drug Immedi- Delayed Immedi- Delayed ate traate treattreat- mntreati- ment metment men Gentamicin sulfate 0.02 0.05 0.03 >0.50 Neomycin sulfate 0.02 0.22 0.33 52.00 Pplymyxii B sul- 0.02 0.02 i fate b a Zinc bacitracin 10.30 1>2.00 1 cacy, it was possible that residual drug in wound washings or the systemic activity of a topically applied drug might account for the efficacy observed. To investigate these possibilities, the antibacterial activity of washings and sera from mice treated with high levels of the various drugs was compared with that of washings and sera from placebo-treated and untreated mice. No significant antibacterial activity was demonstrated with either washings or sera from any treated mice; quantitation of the two test organisms after various periods of exposure to the test or control washings and sera yielded identical results. Since the P. aeruginosa infection is lethal when the infected wound is kept hydrated (7), it seemed possible to exploit this fact and use survival as a measure of efficacy in the evaluation of topical antibacterial agents that possess antipseudomonad activity in vitro. To test this possibility, several commercial topical agents were evaluated against the P. aeruginosa wound infection (Table 5). The two combination products and the gentamicin formulation were quite efficacious in the infection, producing >80% survival, as compared with 13% survival in placebo-treated controls. The triclobisonium ointment demonstrated slight activity and the nitrofurazone preparation was inactive in the model infection. As a firther means of comparing the efficacy of topical antibacterial agents in protecting mice against the lethal effect ofap. aeruginosa infection, the PD." was determined for several of the agents. The results of a typical experiment are presented in Table 6. Polymyxin B and gentamicin were quite active, exhibiting PD., values of <0.1%. Neomycin was also active but to a lesser extent than either polymyxin B or gentamicin. Nitrofurazone, as could

VOL. 10, 1976 TABLz 5. Effect of commercial topical products on survival in mice with surgical wounds infected with P. aeruginosa Formulation % Survivala Gentamicin cream 95.3 ± 3.1 Polymyxin B-bacitracin-neomycin 92.6 ± 5.3 ointment Polymyxin B-bacitracin ointment 84.0 ± 6.0 Triclobisonium chloride ointment 36.7 ± 8.3 Nitrofurazone cream 9.5 ± 0.5 Untreated control 13.0 ± 3.2 a Mean ± standard error of two to four experiments. TABLE 6. PD,0 of selected antibacterial agents in mice infected with P. aeruginosa Antibacterial agent PD5 (%) Polymyxin B sulfate 0.03 Gentamicin sulfate 0.05 Neomycin sulfate 0.13 Nitrofurazone >1.00 be expected from its lack of activity in vitro and in vivo against P. aeruginosa in earlier studies, was not active at the concentrations tested. DISCUSSION Although there is some controversy as to the effectiveness of topical chemotherapy in certain types of local infections, there are several clinical situations in which topical therapy is apparently beneficial (1, 2, 5, 7). Topical antimicrobial formulations are widely used in the field of dermatology (9). Because problems related to the development of resistance, sensitization, toxicity, and other undesirable effects have been encountered with several of the substances commonly used topically, the search has continued for new and more efficacious substances that can be utilized as topical chemotherapeutic agents. The model infections described herein have been established for use as a screening tool for identifying and evaluating substances that may be employed as topical antibacterial agents. By utilizing standardized materials and methods, we have sought to eliminate some of the variables associated with the evaluation of topical antibacterial formulations, with the ultimate aim of bringing the most efficacious candidates to clinical trials in which the final assessment of efficacy must be made. The primary purpose of this study was to evaluate the response of the model infections to treatment with various topical antimicrobial agents and preparations, most of which are employed in the clinic. Since topical agents TOPICAL CHEMOTHERAPY 49 may be either synthetic compounds or antibiotics, representatives of both types were included in this study. The results obtained indicate that the infections are quite responsive to treatment with efficacious topical agents of various potencies. In addition, various levels of efficacy could be detected, as determined by inhibition or reduction of bacterial wound counts and, in the case of the P. aeruginosa infection, by the survival of lethally infected mice. Furthermore, the responsiveness of the experimental infections to treatment with various concentrations of the commercial topical agents allowed the determination of ED50. Although an ED,0 was arbitrarily defined, its determination permits a more definitive and valid comparison of the activities of various agents than does the use of single concentrations of agents. In the case of the P. aeruginosa infection, the determination of PDo also aids in the evaluation of the topical efficacy of test substances. Results obtained from the screening of newly synthesized compounds indicate that the model infections are indeed useful in assessing the topical efficacy of potential agents (unpublished data). For most of the topical agents employed in this study, there appeared to be some correlation between activities in vitro and in vivo. Agents that exhibited low MICs against a particular test organism in vitro usually demonstrated good efficacy when the immediate (prophylactic)-treatment schedule was followed for the experimental infection with that same organism; conversely, agents inactive in vitro showed little or no efficacy in vivo. The correlation was also observed with MBCs. Agents that demonstrated low MBCs in vitro were found to be efficacious when the delayed (therapeutic)- treatment schedule was followed; agents with high MBCs usually showed no activity in vivo. Thus, to some extent, the efficacy of an agent could be predicted from its activity in vitro. Our experience to date in evaluating new compounds has revealed several potential agents that exhibited marked activity in vitro but were relatively ineffective in the experimental infections (unpublished data). Other investigators have made similar observations (4). Although both evaluations, in vitro and in vivo, involve direct contact between the antibacterial agent and the microorganism, the conditions in the infected wound are more complex than those found in the test tube and may affect adversely the activity of an antibacterial agent. For example, an agent markedly active in vitro may be inactivated or neutralized in the wound by the presence of serum, tissue fluids, enzymes, or other substances not present in the

50 McRIPLEY AND WHITNEY test tube. Thus, potential topical agents must be evaluated in vivo under simulated clinical conditions before their true topical efficacy can be ascertained. One criticism commonly leveled when topical chemotherapy is studied in animals is that the topically administered drug has probably been ingested by the animals. Obviously, this can be a serious problem, but in our experience it is only occasionally encountered. For example, in this study, only mice treated with the commercial gentamicin formulation were observed ingesting it from the backs of cagemates. Because the ingestion of agents that are active when administered orally can interfere drastically with the evaluation of the substance as a topical agent, measures must be taken to prevent such ingestion. In the case of the gentamicin formulation, the problem was resolved by isolating the treated mice in small individual cages in which movement was severely restricted. Other means of preventing ingestion are the use of restrainers, collars, or coverings for the treatment site or the addition to the formulation of some inert chemical that discourages ingestion. It should be emphasized that the problem of ingestion occurs only infrequently with most of the commonly used topical preparations and vehicles. Other criticisms sometimes raised when the efficacy of a topical formulation is evaluated culturally are that the activity observed may be due to residual drug and drug carry-over or to systemic activity resulting from topical absorption. In some instances these criticisms may be valid. Therefore, in evaluating topical chemotherapy on a cultural basis, it is necessary to demonstrate that these factors are not involved in the efficacy produced. Simple measures such as those employed in this study can be used to eliminate the possibility that drug carry-over and systemic activity have participated in the cultural results. In the case of residual drug and drug carry-over, there are various preventive measures that can be taken to reduce or abolish their involvement in the cultural evaluation. For example, substantial amounts of residual formulation should be removed prior to culturing a lesion, there should be a definite interval between the termination of therapy and the beginning of cultural evaluations, and the cultural procedures should employ several dilutions before plating. Admittedly, the quantity of formulation applied to each wound as part of the experimental procedure is somewhat large, being analogous ANTIMICROB. AGENTS CHZMOTHZ]t. to applying 1.5 kg of formulation on a 77-kg human. However, we have found that it is necessary to apply a formulation in the experimental wound rather heavily in order to obtain a therapeutic response. Since the mice are caged in groups of 10, there is, obviously, some loss of formulation from the treated site. But, as the data show, the variation in bacterial wound counts between mice in a group is less than might be expected considering the conditions employed. It should be noted that antimicrobial agents formulated for topical usage, in contrast to parenterally and orally administered agents, are often used at the same concentration in both animals and humans. The amount of drug applied and the frequency of treatment are usually determined by the size, type, and nature of the superficial infection being treated and not by the species of the host. Thus, in this study commercial topical antimicrobial preparations, intended for human use, were applied "as is" on the infected surgical wounds. In addition, freshly prepared formulations were made with the antimicrobial agent incorporated within the range of the "use concentration." The surgical-wound model, in addition to its use in the evaluation of the topical antibacterial activity of test substances, has several other applications. For example, a subjective evaluation of the effect of a particular substance on wound healing can be made by examining the gross appearance of the wound at various intervals after treatment. Moreover, the application of the test material to the denuded suface of the wound can reveal topical absorption and toxicity. Finally, the model infections can be used to evaluate the activity of a substance that has been administered parenterally (unpublished data). Although the surgical-wound model can provide useful and valuable information about the effect of topical treatment in mice, as mentioned previously, the ultimate determination of topical efficacy and safety must be made in humans. ACKNOWLEDGMENTS We are indebted to H. Basch for MIC and MBC determinations, to D. Hsieh and R. Schwind for help in animal studies, to R. George and A. Hellwig for technical assistance, and to H. Gadebusch and D. Frost for critical review of the manuscript. LITERATURE CITED 1. Beizer, F. O., 0. Salvatierra, Jr., R. T. Schweizer, and S. L. Kountz. 1973. Prevention ofwound infections by topical antibiotics in high risk patients. Am. J. Surg. 126:180-185. 2. Bingham, R., W. H. Fleenor, and S. Church. 1974. The local use of antibiotics to prevent wound infection. Clin. Orthop. Relat. Res. 99:194-200.

VOL. 10, 1976 TOPICAL CHEMOTHERAPY 51 3. Gadebusch, H. H., and H. I. Baich. 1974. A new antimicrobial nitrofuran, tran.-5-amino-3[2-(5-nitro-2-furyl)vinylj 42-1,2,4-oxadiazole: antibacterial, antifungal, and antiprotozoan activity in vitro. Antimicrob. Agents Chemother. 6:263-267. 4. Grunberg, E., and R. J. Schnitzer. 1959. Experimental approach to topical antibacterial therapy. Ann. N.Y. Acad. Sci. 82:114-118. 5. Hummel, R. P., B. G. MacMilln, and W. A. Altemeier. 1970. Topical and systemic antibacterial agents in the treatment of burns. Ann. Surg. 172:370-384. 6. McRipley, R. J., and R. R. Whitney. 1976. Characterization and quantitation of experimental surgicalwound infections used to evaluate topical antibacterial agents. Antimicrob. Agents Chemother. 10:38-44. 7. Monenef, J. A. 1973. Burns. N. Engl. J. Med. 288:444-454. 8. Reed, L. J., and H. Muench. 1938. A simple method of estimating fifty percent end points. Am. J. Hyg. 27:493-499. 9. Sauer, G. S. 1973. Manual ofskin diseases. J. B. Lippincott Co., Philadelphia. Downloaded from http://aac.asm.org/ on September 22, 2018 by guest