Quantitative Study of Antibiotic-Induced Susceptibility to

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, July 199, p. 1348-1353 66-484/9/71348-6$2./ Copyright 199, American Society for Microbiology Vol. 34, No. 7 Quantitative Study of Antibiotic-Induced Susceptibility to Clostridium difficile Enterocecitis in Hamsters H. E. LARSON* AND S. P. BORRIELLO Microbial Pathogenicity Research Group, Division of Communicable Diseases, Clinical Research Centre, Watford Road, Harrow, Middlesex, England HAI 3UJ Received 27 September 1989/Accepted 13 April 199 Commonly used antibiotics were compared for their ability to induce Clostridium difficile enterocecitis and death in hamsters. Susceptibility to infection with C. difficile was measured by calculating 5% lethal doses (in CFU) for hamsters for various intervals after antibiotic treatment. Infection occurred after very small doses of C. difficile were given to hamsters treated with clindamycin, ampicillin, flucloxacillin, and cefuroxime; there was little difference between the antibiotics in the degree of susceptibility that they induced. A large difference in the duration of susceptibility was observed, however, with susceptibility being temporary following ampicillin, flucloxacillin, and cefuroxime administration but long-lived following clindamycin administration. A larger dose of ampicillin, multiple doses of ampicillin, and a combination of antibiotics had comparatively small effects on the duration of susceptibility. C. dfficile growth and toxin production in in vitro suspensions of cecal contents were found to correlate closely with in vivo hamster infectivity. A persisting loss of colonization resistance following antibiotic treatment may be a type of postantibiotic effect. Although these results cannot be applied directly to humans, they suggest lines of further investigation into how antibiotics may differ in producing risks of C. difficile infection and pseudomembranous colitis in patients. Prior to the recognition that pseudomembranous colitis (PMC) was caused by Clostridium difficile, it was known to be a risk of antibiotic treatment, especially lincomycin and clindamycin (11). Discovery of the microbial pathogenesis of PMC (18) has permitted a more careful assessment of the relationship between antibiotic treatment and C. difficile infection (17). A large number of antibiotics have been found to facilitate infection in both animals and humans (2, 1). There has been suspicion that some antibiotics are more likely to result in C. difficile-associated disease than others. Aronsson et al. (1) compared the relative risk of antibiotics by determining the number of C. difficile toxin-positive cases relative to antibiotic sales in Sweden. In contrast, Pierce et al. (19) failed to find differences in the rates of antibioticassociated colitis after the administration of several antibiotics, although the power of that study to detect significant differences likely was small. Quantitative risk estimations in humans are bedevilled by problems of sample size, uncertainties about how much and how often antibiotics are actually consumed, differences in routes of administration, use of combination antibiotic therapies, and ignorance about whether individuals have actually been exposed to C. diffcile in their environments. We undertook an experimental quantitative comparison of various antibiotics using both in vivo and in vitro methods. It is known that oral vancomycin treatment makes hamsters susceptible to infection with very small doses of C. difficile (18). We determined the 5% lethal dose (LD5) for hamsters of C. difficile spores after administration of single oral doses of various antibiotics and assessed the capacity of suspensions of hamster cecal contents to support growth of C. difficile in vitro. * Corresponding author. MATERIALS AND METHODS C. difficile. Two strains (B-1 and H-1) of C. difficile isolated from patients with PMC and known to produce enterocecitis when administered to clindamycin-treated hamsters (17) were used in this study. Both strains produce toxins A and B. Antibiotic MICs were determined by a microdilution technique. For strain B-1 the MICs were as follows: clindamycin, 256,ug/ml; ampicillin,.5,ug/ml; flucloxacillin, 4,ug/ml; and cefuroxime, 128,ug/ml. Strain H-1 was inhibited by.13,ug of clindamycin per ml and.5,ug of ampicillin per ml. For inoculation of hamsters, a spore preparation was made from a 5-day Robertson cooked meat medium culture by using alcohol shock. The culture broth was centrifuged at 2, x g for 1 min, suspended in 5 ml of brain heart infusion (Difco Laboratories, Detroit, Mich.) with cysteine hydrochloride (BHIC;.5% [wt/vol]), mixed with 5 ml of 1% ethanol, held for 1 h at ambient temperature, and centrifuged again as described above. The pellet was washed once in BHIC and suspended in 5 ml of BHIC, and serial 1-fold dilutions in BHIC were prepared. The number of viable spores in the inocula was estimated by seeding.5 ml of diluted suspension on a blood agar plate (Difco), incubating it anaerobically at 37 C for 48 h in an anaerobic cabinet (Forma Scientific, Marietta, Ohio), and recording the number of colonies. Animals. Syrian hamsters obtained from breeding colonies at the National Institute for Medical Research, London, England, were individually maintained in sterilized filterlidded polycarbonate boxes (Stephen Clark Fabrications Ltd., Alva, Scotland) and were given autoclaved bedding, feed, and tap water (17). Wire cage floors were used to prevent coprophagia in one experiment. Antibiotic solutions and suspensions of C. difficile were administered orally in.25-ml volumes from the end of a 1-ml syringe. Antibiotics, as supplied for clinical use, were diluted in distilled water to yield 3 mg in.25 ml. All manipulations of antibiotic-treated animals were carried out with attention given to the preven- 1348 Downloaded from http://aac.asm.org/ on September 14, 218 by guest

VOL. 34, 199 QUANTITATIVE STUDY OF SUSCEPTIBILITY TO C. DIFFICILE 1349 tion of cross-contamination of infection from the environment of the animal holding unit. Control antibiotic-treated hamsters were inoculated with sterile broth after the inoculations of C. difficile were completed. Observations were continued for 7 days, at which time surviving animals were sacrificed. All animals were examined for evidence of cecitis İnfectivity calculation. Susceptibility of hamsters to C. difficile enterocecitis was measured by calculating the number of CFU required to produce enterocecitis in half of the animals inoculated with the highest dilution that produced disease. First, the LD5 was determined by challenging hamsters with log1o dilutions of a spore preparation after they were given a single oral dose of antibiotic. Three hamsters were challenged per dilution and observed as described above. The number of hamsters with enterocecitis at each dilution was recorded, and the Karber method of endpoint calculation was used to yield the LD5 (12). Second, the CFU in each dilution of the spore suspension was estimated as described above. The CFU, as determined by the estimated number of C. difficile at the endpoint dilution, was divided by the LD5, and this value was rounded to give the nearest whole CFU per 1 LD5. To conserve animals but still give a complete picture of the duration of susceptibility after various experimental variables, some challenges were performed with only a single large dose (14 CFU) of spores. Intervals after antibiotic treatment were selected for challenge that defined the duration of susceptibility, in days. In vitro suspensions. Suspensions of cecal contents for in vitro studies were prepared from freshly sacrificed hamsters who had received antibiotics but not C. difficile, as described previously (6). Briefly, cecal contents were removed from each of two hamsters for each day following antibiotic treatment, suspended to 1%o (wt/vol) in prereduced distilled water, and placed in bijoux in duplicate aliquots of 5. ml each. Logarithmic-growth-phase C. difficile was prepared by inoculating an overnight culture into fresh BHIC and incubating it anaerobically for 6 h at 37 C. This was centrifuged, washed in 2 ml of fresh broth, centrifuged, and suspended in 1 ml of fresh broth. The suspension was diluted 1:1, and 5 p.l was inoculated into each suspension of cecal contents. This inoculum typically contained 5 x 16 CFU. Suspensions of cecal contents were then incubated at 37 C anaerobically and examined for C. difficile growth at 48 h by a micromethod as described elsewhere (6). To test for toxin production, suspensions of cecal contents were centrifuged and supernatants were serially diluted twofold. One hundred microliters of each dilution was added to confluent monolayers of Vero cells in microdilution plates and observed at 24 and 48 h for evidence of a C. difficile toxin B cytopathic effect. To test for the possible presence of residual antibiotic, suspensions of cecal contents were centrifuged at 12,9 x g for 3 min, and supernatants were filtered through.45-,um-pore-size filters. Filtrates were inoculated, incubated, and tested for growth of C. difficile and toxin production in the same manner as described previously for suspensions of cecal contents (6). Some suspensions of cecal contents were assayed for clindamycin by the method of Roberts et al. (2), for ampicillin by using Sarcina lutea as the test organism, and for P-lactamase by the method of Rolfe and Finegold (21). TABLE 1. Calculation of LD5 of C. difficile B-1 2 days after a single 3-mg dose of clindamycin Dilution of C. difficile No. with enterocecitis/ No. of CFU/ spore suspension no. inoculated.5 ml inoculated 1-2 3/3 251 lo-3 3/3 41 lo-4 3/3 5 1-5 1/3 1 1-6 /3 1-7 /3 Broth only /2 RESULTS Hamsters given antibiotics followed by broth and those hamsters in the holding stocks that did not receive antibiotics remained well. Table 1 records the results of a sample experiment used to determine susceptibility to C. difficile B-1 2 days after a single oral, 3-mg dose of clindamycin. The LD5 per.25 ml of undiluted challenge inoculum was 4.8 (log1o). There was 4.7 CFU (log1o) in.25 ml of undiluted challenge inoculum. This yielded.8, or approximately 1, CFU/LD5. Table 2 summarizes the results of a series of similar experiments conducted at various intervals between antibiotic treatment and challenge with C. difficile. For up to 2 days after treatment, less than 1 CFU was needed to produce enterocecitis and death for all antibiotics listed in Table 2. After clindamycin treatment, hamsters remained susceptible to small doses of C. difficile for at least 14 days; these animals were maintained on wire cage floors. In contrast, animals treated with ampicillin and cefuroxime did not remain susceptible and were resistant to large doses of C. difficile by 3 days after antibiotic treatment. After flucloxacillin treatment, hamsters were still susceptible at 3 days. In one experiment, suspensions of C. difficile were administered immediately following the administration of ampicillin (Table 2, day interval). This produced susceptibility to 6 CFU of C. difficile. Since it appeared that antibiotics were distinguished by the duration rather than by the degree of susceptibility they produced, further experiments were performed by challenging the hamsters with a single large inoculation of C. difficile spores. Table 3 summarizes experiments which further compared various antibiotic regimens by challenging the hamsters with 14 spores at various intervals after antibiotic treatment was completed. Clindamycin-treated animals were still susceptible 74 days after a single 3-mg dose; these hamsters were not maintained on wire cage floors. An ampicillin dose of.3 mg did not produce susceptibility, whereas a dose of 3 mg produced susceptibility that lasted 2 days with evidence of recovery of resistance by 3 days. A TABLE 2. LD5 of C. difficile B-1 at various intervals after a single 3-mg dose of antibiotic Antibiotic LD5 (CFU) at the following intervals (days) 1 2 3 5 14 Clindamycin 1 2 Ampicillin 6 8 >1 x 13 >2 x 13 Flucloxacillin 5 6 Cefuroxime 7 >1 x 14 Downloaded from http://aac.asm.org/ on September 14, 218 by guest

135 LARSON AND BORRIELLO ANTIMICROB. AGENTS CHEMOTHER. TABLE 3. Susceptibility to C. difficile enterocecitis at various intervals after various doses of antibiotic treatment No. dying/no. inoculated at the following intervals (days) between Antibiotic Dose (mg) No. doses of antibiotic treatment and challenge: 1 2 3 4 5 6 7 8 11 16 74 Clindamycin 3. 1 3/3 3/3 Ampicillin.3 1 /4 3. 1 4/4 4/4 1/4 /3 3. 1 7/8 /8 /8 /8 3. 3 3/4 1/9 4/5 1/4 /4 /4 Flucloxacillin 3. 1 6/6 7/8 2/8 3. 3 4/4 2/4 3/4 1/4 /4 /4 Flucloxacillin + ampicillin 3. 1 3/4 /4 /4 1-fold increase in the dose of ampicillin from 3 to 3 mg extended the period of susceptibility by only 1 day (seven of eight animals), and hamsters regained resistance to C. difficile by day 4. A single daily dose of 3 mg of ampicillin for 3 successive days produced an extended, although occasionally variable, susceptibility compared with those produced by the other regimens. By day 6 the animals were resistant to challenge. Flucloxacillin treatment produced results similar to those obtained with 3 mg of ampicillin, in that susceptibility persisted for 3 days. Flucloxacillin for 3 successive days, as with ampicillin, extended the period of susceptibility by 3 days. A fixed-dose combination of flucloxacillin and ampicillin did not increase the duration of susceptibility produced by flucloxacillin alone. To assess the role of clindamycin susceptibility in the development of enterocecitis, clindamycin-treated hamsters were challenged with a strain of C. difficile that was susceptible to clindamycin, strain H-1. Hamsters were resistant to challenge with more than 1i C. difficile spores 3 days after clindamycin treatment but were susceptible to 21 CFU 26 days after treatment. This degree of susceptibility was 1-fold less than that to C. difficile B-1. To study the correlation between the duration of susceptibility and antibiotic persistence, cecal contents of antibiotic-treated hamsters were assayed for antibiotic activity at various intervals after administration of a single 3-mg dose (Table 4). Clindamycin was detected up to 11 days after treatment. Concentrations at that time would have been sufficient to inhibit C. difficile H-1. In contrast, ampicillin was detected in low concentrations in two of five animals 1 day after treatment and not thereafter. P-Lactamase activity was detected in a single animal 3 days after treatment with ampicillin. Table 5 records the in vitro changes in numbers of C. difficile and toxin titers in suspensions' of cecal contents prepared from hamsters given antibiotics. The presence of toxin in the suspensions correlated significantly with growth (analysis of variance, P <.1, r =.87). There was no instance in which growth occurred but toxin was not produced. Suspensions from animals given.3 mg of ampicillin permitted neither C. difficile growth nor appreciable toxin production. This was not the result of residual antibiotic since filtrates of the suspensions all permitted growth and toxin production. Suspensions of cecal contents from animals given 3. mg of ampicillin were not inhibitory and permitted toxin production on days 1 and 2 after treatment but were inhibitory to growth and toxin production by day 3. Cefuroxime treatment resulted in stationary numbers without toxin production on day 1 and inhibitory suspensions of cecal contents on subsequent days. Ampicillin (3 mg) and flucloxacillin treatment resulted in suspensions of cecal contents that allowed growth and toxin production for a duration that was clearly longer than that after treatment with 3 mg of ampicillin. Some suspensions of cecal contents appeared to show partial effects either in growth or inhibition, and there were a few instances in which growth was inhibited but toxin was produced. Such findings are the result of averaging results for preparations from duplicate animals in which some suspensions of cecal contents were inhibitory and some permitted growth. Such mixed results were also evident in vivo, e.g., Table 3, days 3 through 6 after ampicillin treatment and days 3 through 7 after flucloxacillin treatment. DISCUSSION Although there is an extensive amount of information in the literature on the capacity of various antibiotics to induce C. difficile enterocecitis in hamsters (e.g., see references 2 TABLE 4. Assay for antibiotic activity in wet cecal contents at various intervals after a single 3-mg-dose Time (days) Clindamycin Ampicillin after Animal Concn Animal Concn activity treatment no. (Lg/g) no. (4/g) 1 1 61 1 <.4 2 91 2.8 3 <.4 4 <.4 5.8 2 1 35 1 <.4 2 28 2 <.4 3 27 3 1 <.4 2 <.4 + 8 1 9 11 1 4 2 6 19 1 <.5 2 <.5 21 1 <.5 2 <.5 Downloaded from http://aac.asm.org/ on September 14, 218 by guest

() * It l+i+ I+1I+ 1+ A -- 1.~1 t,4~~~~~-- '- C) ~ ~Di ~3 -j :I :PWO 'oo'ob 1+ R1+1+1+1+1 C o LQLaOb w z D w A6 " ") z 'I *N 1+ *- *. cl p o) oo) C t'i4 t - e 1+ 1+1+1+1+ 1+ z Z z 3 co oo 1+ **oi7 _) -3 It C) : + co _.D b3 i13:x bo _Q.. C Q *-I CI. DD B. z c c -. n (I (D DQ r-l 1v co DI. co. C) ai r md. (D, -. '. Di DiD - Xi _ DiD ' DiD U' (D F and 1), we are not aware of previous efforts to measure susceptibility or to use characterized challenge strains. By and large, previous investigators have allowed antibiotictreated animals to become infected fortuitously, although no doubt, with time, the environments of animal holding units become heavily contaminated with a predominant strain of C. difficile. The animal containment system we used prevents infection from ambient sources and permits challenge experiments with measured numbers of defined microorganisms at specified times. An earlier study, in which containment methods were used, showed that hamsters were susceptibile to a very small number of C. difficile after prolonged oral vancomycin treatment, whereas untreated hamsters were uniformly resistant to challenges with large doses of C. difficile (18). The current study considerably extended this approach and permits a quantitative measure of susceptibility after treatment with several commonly used antibiotics. The main findings are that all antibiotics tested produced susceptibility to very small doses of C. difficile but that these antibiotics differed to a smaller or greater extent in the duration of susceptibility after treatment. A 1-fold increase in the dose or the administration of three doses had a comparatively small effect on the duration of susceptibility. Persistence of susceptibility was not necessarily due to continued detectability of the antibiotic. Finally, the state of susceptibility was modeled in vitro, and the correlation with susceptibility in the animal was very strong. An effect on colonization resistance that outlasts detectable antibiotic activity may be a type of postantibiotic effect. Both clindamycin and ampicillin are associated with postantibiotic effects in other systems (8); clindamycin is known to produce the more prolonged effects and to act this way in vivo as well as in vitro. Prolonged excretion of active drug, active metabolites, or both may produce an especially strong effect. The mechanisms of the postantibiotic effect, and particularly, how they might apply in the gut, are not immediately obvious. A "postantibiotic effect" from clindamycin may simply be the elimination of the particular species responsible for colonization resistance. The data in Table 2 support the idea that susceptibility is an all-or-nothing state; reversion to resistance occurred rapidly. At some intervals after antibiotic treatment, individual differences in susceptibility were noted, especially in animals which had received multiple doses of ampicillin or flucloxacillin. It is not clear from the results of these experiments why individual animals might vary in their responses to multiple antibiotic doses. It is impossible from these experiments to say whether the apparent small differences in CFU/LD5O between clindamycin (1 and 2, Table 2) and the other antibiotics tested (5 to 8, Table 2) are real. Residual clindamycin activity in the gut appeared to inhibit infection with clindamycin-susceptible strain H-1. However, this strain had a slightly reduced capacity to infect hamsters even when clindamycin activity was undetectable. Clindamycin-susceptible strains of C. difficile are fully capable of causing illness in humans who receive clindamycin (9), but the onset of infection may be delayed until levels of antibiotic in the gut have abated. The occurrence of infection after treatment with ampicillin, to which all virulent strains are highly susceptible, has been ascribed to the induction of,b-lactamase in cecal contents (21). However, we found that small doses of C. difficile spores were able to induce disease immediately after a single oral dose of ampicillin, prior to the appearance of,b-lactamase. Overall, there seems to be no direct correlation between susceptibility to an antibiotic and Downloaded from http://aac.asm.org/ on September 14, 218 by guest VOL. 34, 199 QUANTITATIVE STUDY OF SUSCEPTIBILITY TO C. DIFFICILE 1351

1352 LARSON AND BORRIELLO the capacity of a C. difficile strain to produce disease in either hamsters or humans (9). Borriello and Barclay (6) have shown in a clinical study that the growth of C. difficile in fecal suspensions in vitro parallels in vivo colonization. The correlation between the experiments reported in Tables 3 and 5 is striking. This suggests that the capacity of C. difficile to replicate in the fecal stream is a fundamental feature of the pathogenesis of the infection. The mechanism of resistance to colonization by C. difficile in untreated animals postulated by Borriello and Barclay (5, 6) requires the presence of viable microorganisms; it does not appear to be mediated by preformed bacterial products or metabolites that can be detected in fecal filtrates (5, 6). The very rapid onset of susceptibility noted in the present study is consistent with a requirement for viable organisms rather than soluble substances. Restoration of resistance after antibiotic treatment would then be the result of their regrowth. Mediation of resistance by means of viable microorganisms explains the apparently successful efforts to manage relapses of PMC by administering fecal suspensions by enema (7, 22). How might these results be applied to clinical situations? Differences between hamsters and humans limit the clinical applicability of our results. One difference concerns species variation in the way in which clindamycin is metabolized. In 15 patients given clindamycin intravenously (14) for 48 h perioperatively, levels were still present in feces and, indeed, were still increasing at day 5. Ten adult volunteers given oral clindamycin for 7 days (13) had substantial levels in feces 2 days after they concluded the course of therapy, but there was no detectable antibiotic activity at 9 days. The review of clindamycin by Klainer (16) emphasizes its fecal persistence and the potential role of this characteristic in the development of antibiotic-associated colitis. We found that clindamycin is detectable in hamsters 11 days after a single dose but not at 19 days. The human data suggest some persistence of the drug in feces, but not for as long as that which we observed in hamsters. A second difference concerns the normal gut flora in hamsters and humans responsible for excluding C. difficile. It is not known what specific microorganisms are responsible for colonization resistance in either hamsters or humans, even at the genus level (4). However, we have successfully restored hamster colonization resistance with a suspension of human faeces (H. E. Larson, unpublished data). The relevant microorganisms and susceptibilities may not be different in the two species, but this possibility needs to be confirmed. A third difference concerns coprophagia. Prevention of coprophagia did not shorten the duration of susceptibility after clindamycin treatment. We cannot be sure it did not hasten the return of colonization resistance after treatment with the other antibiotics. However, hamsters remained susceptible to clindamycin even when coprophagia was allowed. A fourth difference concerns the many more permutations of dose, frequency, and route of administration which are features of clinical use and which were not tested by our methods. Our results predict that differences in dose and frequency of administration are less important than the type of antibiotic, but this prediction might be overcome by very large doses or very long courses of treatment. One of the most striking of our findings was the very low infective dose of C. difficile required to initiate infection. After clindamycin treatment, we reproducibly found that only one or two microorganisms were needed. It is not known what dose of C. difficile is required to infect human subjects. It may be very low since nosocomial C. difficile ANTIMICROB. AGENTS CHEMOTHER. infections appear to occur as a consequence of exposure to contaminated fomites rather than to contaminated food or water. This is likely to mean ingestion of only a few microorganisms. We are not aware that the duration of susceptibility to C. difficile infection has previously been assessed as a risk factor for PMC. Prolonged susceptibility after antibiotic treatment offers more opportunity for exposure to and infection with C. difficile. It has not usually been possible to correlate the risk of C. difficile colitis with dose or duration of antibiotic treatment. This difficulty is most likely due to overriding differences in the degree to which at-risk patients have been exposed to environmental sources of C. difficile. Given the low rates of colonization in normal subjects, exposure to the organism is likely to be the single most important factor after antibiotic treatment that determines the risk of disease. The risk of exposure, however, is clearly related to the duration of susceptibility. PMC has been shown to occur after even short courses of cefoxitin given as perioperative infection prophylaxis (3). However, the postoperative duration of prophylactic antibiotic regimens is controversial, and PMC is their most important complication (15). We speculate that shorter rather than longer courses of antibiotic treatment could reduce the risk of complicating antibiotic-associated colitis. ACKNOWLEDGMENTS We thank A. Welch, F. Barclay, and H. Anderson for technical assistance and E. Goodband for typing the manuscript. LITERATURE CITED 1. Aronsson, A., R. MolHby, and C. E. Nord. 1982. Clostridium difficile and antibiotic associated diarrhoea in Sweden. Scand. J. Infect. Dis. 35(Suppl.):53-58. 2. Bartlett, J. G., T. W. Chang, N. Moon, and A. B. Onderdonk. 1978. Antibiotic-induced lethal enterocolitis in hamsters: studies with eleven agents and evidence to support the pathogenic role of toxin-producing clostridia. Am. J. Vet. Res. 39:1525-153. 3. Block, B. S., L. J. Mercer, M. A. Ismail, and A. H. Moawad. 1985. Clostridium difficile-associated diarrhea follows perioperative prophylaxis with cefoxitin. Am. J. Obstet. Gynecol. 153:835-838. 4. Borriello, S. P. 1989. Influence of the normal flora of the gut on Clostridium difficile, p. 239-251. In R. Grubb, T. Midtvedt, and E. Norin (ed.), The regulatory and protective role of the normal microflora. Macmillan, London. 5. Borriello, S. P., and F. E. Barclay. 1984. Colonization resistance to Clostridium difficile infection. Microecol. Ther. 14:75-87. 6. Borriello, S. P., and F. E. Barclay. 1986. An in-vitro model of colonisation resistance to Clostridium difficile infection. J. Med. Microbiol. 21:299-39. 7. Bowden, T. A., A. R. Mansberger, and L. E. Lykins. 1981. Pseudomembranous enterocolitis: mechanism of restoring floral homeostasis. Am. Surg. 47:178-183. 8. Craig, W. A., and B. Vogelman. 1987. The postantibiotic effect. Ann. Intern. Med. 16:9-92. 9. Dzink, J., and J. G. Bartlett. 198. In vitro susceptibility of Clostridium difficile isolates from patients with antibiotic-associated diarrhea or colitis. Antimicrob. Agents Chemother. 17: 695-698. 1. Fekety, R., J. Silva, R. Toshniwal, M. Allo, J. Armstrong, R. Browne, J. Ebright, and G. Rifkin. 1979. Antibiotic-associated colitis: effects of antibiotics on Clostridium difficile and the disease in hamsters. Rev. Infect. Dis. 1:386-396. 11. Gorbach, S. L., and J. G. Bartlett. 1977. Pseudomembranous enterocolitis: a review of its diverse forms. J. Infect. Dis. 135:S89-S94. 12. Hawkes, R. A. 1979. General principles underlying laboratory diagnosis of viral infections, p. 32-35. In E. H. Lennette and Downloaded from http://aac.asm.org/ on September 14, 218 by guest

VOL. 34, 199 QUANTITATIVE STUDY OF SUSCEPTIBILITY TO C. DIFFICILE 1353 N. J. Schmidt (ed.), Diagnostic procedures for viral, rickettsial and chlamydial infections, 5th ed. American Public Health Association, Washington, D.C. 13. Heimdahl, A., and C. E. Nord. 1982. Effect of erythromycin and clindamycin on the indigenous human anaerobic flora and new colonization of the gastrointestinal tract. Eur. J. Clin. Microbiol. 1:38-48. 14. Kager, L., L. Liljeqvist, A. S. Mahnborg, and C. E. Nord. 1981. Effect of clindamycin prophylaxis on the colonic microflora in patients undergoing colorectal surgery. Antimicrob. Agents Chemother. 2:736-74. 15. Kaiser, A. B. 1986. Antimicrobial prophylaxis in surgery. N. Engl. J. Med. 315:1129-1138. 16. Klainer, A. S. 1987. Clindamycin. Med. Clin. North Am. 71:1169-1175. 17. Larson, H. E., A. B. Price, and S. P. Borriello. 198. Epidemiology of experimental enterocecitis due to Clostridium difficile. J. Infect. Dis. 142:48-413. 18. Larson, H. E., A. B. Price, P. Honour, and S. P. Borriello. 1978. Clostridium difficile and the aetiology of pseudomembranous colitis. Lancet i:163-166. 19. Pierce, P. F., R. Wilson, J. Silva, Jr., V. F. Garagusi, G. D. Rifkin, R. Fekety,. Nunez-Montiel, Jr., V. R. Dowell, and J. M. Hughes. 1982. Antibiotic-associated pseudomembranous colitis: an epidemiologic investigation of a cluster of cases. J. Infect. Dis. 145:269-274. 2. Roberts, A. P., J. B. Eastwood, P. E. Gower, C. M. Fenton, and J. R. Curtis. 1978. Serum and plasma concentrations of clindamycin following a single injection of clindamycin phosphate in maintenance haemodialysis patients and normal subjects. Eur. J. Clin. Pharmacol. 14:435-439. 21. Rolfe, R. D., and S. M. Finegold. 1983. Intestinal P-lactamase activity in ampicillin-induced, Clostridium difficile-associated ileocecitis. J. Infect. Dis. 147:227-235. 22. Schwan, A., S. Sjolin, V. Trottestam, and B. Aronsson. 1984. Relapsing Clostridium difficile enterocolitis cured by rectal infusion of normal faeces. Scand. J. Infect. Dis. 16:211-215. Downloaded from http://aac.asm.org/ on September 14, 218 by guest