High Frequency of Antimicrobial Resistance in Human Fecal Flora

ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Dec. 1988, p. 181-186 66-484188112181-6$2./ Copyright 1988, Americn Society for Microbiology Vol. 32, No. 12 High Frequency of Antimicrobil Resistnce in Humn Fecl Flor STUART B. LEVY, 12* BONNIE MARSHALL,' SUSAN SCHLUEDERBERG,' DEBORAH ROWSE,' AND JOANNE DAVIS1 Deprtments of Moleculr Biology nd Microbiology' nd Medicine,2 Tufts University School of Medicine nd New Englnd Medicl Center, Boston, Msschusetts 2111 Received 25 April 1988/Accepted 9 September 1988 The frequency of resistnce to seven different ntimicrobil gents ws exmined in the erobic grmnegtive gut flor of over 6 individuls from hospitls, from lbortories where ntibiotics were used, nd from urbn nd rurl communities. In mjority (62.5%) of fecl smples from people without recent history of tking ntibiotics, 1% or more of the totl orgnisms were resistnt to t lest one of the ntibiotics. In bout 4% of the smples, resistnce to more thn one drug ws present t this level. More thn one-third of the smples contined resistnt orgnisms comprising 5% or more of the totl flor exmined. Orgnisms with coresistnce to multiple drugs were found frequently. Individuls tking ntibiotics produced more smples with higher proportion (>5%) of resistnt bcteri, nd these smples lso hd significntly greter number of different resistnce determinnts. This extensive study reveled high prevlence of resistnt bcteri in the gut flor of mbultory nd hospitlized individuls whether or not they were tking ntibiotics. There is worldwide concern bout the ppernce of ntibiotic resistnce in common pthogens of hospitl- nd community-cquired infections (4, 9, 2). While selective pressure cused by ntibiotic usge hs been linked to these findings (2), there hve been little dt on the nturl frequency of ntibiotic resistnce genes in the nornil nonpthogenic flor of mbulting nd hospitlized individuls. Such resistnt bcteri would represent constnt pool of resistnce genes potentilly trnsferble, directly or indirectly, to humn pthogens. We nlyzed fecl specimens from humns in the greter Boston, Mss., re for both the presence nd frequency of nturlly occurring erobic grmnegtive bcteri with resistnce to one or more of seven different ntimicrobil gents. We found tht resistnce ws common in the gut flor nd the resistnt strins constituted mjor frction of the totl flor in more thn one-third of the smples. This density of resistnce ws found in smples from individuls who did not ingest ntibiotics s well s from those who did. MATERIALS AND METHODS Description of smple popultions. Over 16-month period (December 1977 through Mrch 1979), we collected nd tested 974 fecl smples from 64 hospitlized nd mbultory humn donors. The popultion consisted of hospitlized s well s norml mbultory people. The ptient smples were obtined from four medicl wrds of the New Englnd Medicl Center Hospitl nd the clinicl study unit. The mbultory popultion consisted of lbortory workers nd urbn nd rurl dwellers. The lbortory donors cme from eight greter Boston reserch nd clinicl lbortories; 1 smples representing seven donors were obtined from lbortory workers t the University of Georgi, Athens. The urbn popultion ws composed of first-yer medicl students nd instructors from the Tufts University nd Hrvrd University medicl schools nd fmily members of lbortory workers. None hd contct with ptients or hospitls. The rurl popultion consisted of fmilies residing in Sherborn, Mss., locted 25 miles west of Boston. Most of * Corresponding uthor. the mbultory contributors nswered questionnire which included the following items: sex, ge, communities of residence nd employment, nd history of ntibiotic therpy within the previous 2 weeks or exposure through working with ntibiotics. The donors were predominntly dult (95.4% over ge 15) with mle/femle distribution of pproximtely.8:1 (Tble 1). The men ge of femles (45 ± 22.1 yers) ws slightly higher thn tht of mles (39.9 ± 2.9 yers); the men ge of the totl popultion studied ws 42.7 ± 21.7 yers, with medin ge of 37 yers (Tble 1). In September 1987, second smller smpling ws performed on group of 8 second-yer Tufts medicl students. The ge rnge of this group ws 22 to 41 yers, with men of 24.5 yers nd medin of 24 yers. No member of this group reported ntibiotic ingestion in the previous 6 months; none hd ptient contct. University guidelines were followed for obtining humn fecl smples. Smple collection nd processing. Fresh fecl smples from hospitlized ptients were collected directly into plstic continer (Lb-Tek; Miles Scientific, Div. Miles Lbortories, Inc., Nperville, Ill.) which ws refrigerted immeditely nd trnsported to the lbortory within 24 h. Ech fecl mss ws smpled internlly by inserting sterile swb into the center. Smples from nonhospitlized individuls were collected by the sme mens or by sterile cottontipped pplictor (Culturette; Scientific Products) inserted into the rectum or directly into fresh feces. Ech Culturette ws immeditely crushed to relese Sturt trnsport medium. Mny individuls were smpled more thn once to evlute the stbility of given gut profile. All individuls provided their own smples nd consented to the use of their excrement for this nlytic study. Except when noted, ll sttistics were bsed on results derived from the first fecl specimen collected from ech donor. The dt on multiple smpling were treted seprtely. Ech swb ws plted onto two pltes of McConkey gr by inoculting one qudrnt of ech plte. The inoculum ws then diluted by two-dimensionl streking onto the remining three qudrnts. All pltes were incubted overnight t 37 C. From ech plte, 5 to 2 clerly isolted colonies 181

182 LEVY ET AL. ANTIMICROB. AGENTS CHEMOTHER. TABLE 1. Study popultion Totl no. of Sex distribution Age (yr) Group individuls b_d smpled Femles" Mlesc Rnge Men + SD" Hospitl ptients 298 16 127 12-95 59.7 ± 17.2 Lbortory workers 13 65 56 19-6 28.9 ± 7.5 Urbn dwellers 19 51 58 1-64 26.5 ± 11.6 Rurl dwellers 62 3 29 1-61 25.4 ± 17.1 Othere 41 29 8 23-63 33.4 ± 14.7 Bsed on the following smple sizes of known ge: hospitl, 297; lbortory workers, 12; urbn dwellers, 19; rurl dwellers, 59; other, 37. b Men ge (includes only those whose sex ws known) ± stndrd devition, 45 ± 22.1 yers; rnge, 1 to 92 yers. ' Men ge (includes only those whose sex ws known) ± stndrd devition, 39.9 ± 2.9 yers; rnge, 1 to 95 yers. d The medin ge for the popultion studied ws 37 yers; the men ge ws 42.7 ± 21.7 yers. e Consists of 11 hospitl stff nd 3 vegetrins. were enumerted nd clssified s lctose fermenters nd lctose nonfermenters. The best of ech pir of McConkey pltes ws selected for replic plting (5) onto series of ntibiotic-contining McConkey pltes (ech contining one of the following ntimicrobil gents: mpicillin, 3,ug/ml; streptomycin, 3,ug/ml; tetrcycline, 1,ug/ml; knmycin, 1 p.g/ml; chlormphenicol, 25,ug/ml; gentmicin, 1,ug/ml; nd nlidixic cid, 3 jig/ml), followed by plte of plin McConkey gr. Smples from the hospitl groups were lso tested with cephlothin, 3 xg/ml. Smples tken in the 1987 survey were tested with the following ntibiotics: mpicillin, streptomycin, tetrcycline, nd knmycin. Trimethoprim t 2,ug/ml in Mueller-Hinton gr ws lso tested. After overnight incubtion t 37 C, the pltes were gin tbulted by comprison with the replic-plted control plte of Mc- Conkey gr lone. Numbers of the vrious colony types resistnt to ech drug were recorded. From control studies, the levels of ntibiotic used nd the size of the colonies recovered excluded the possibility of counting spontneous chromosoml mutnts. The dt obtined from ech smple were converted to n lph-numeric coding system nd nlyzed by computer, using the SPSS (Sttisticl Pckge for the Socil Sciences) progrm (SPSS, Inc., Chicgo, Ill.). The dt were orgnized t two levels in the fecl smple:.1 nd.5% resistnt orgnisms. Sttisticl significnce ws determined by the chi-squre test. RESULTS Frequency of ntibiotic resistnce mong fecl specimens from four different humn popultions. Grm-negtive erobic fecl bcteri from hospitlized individuls, those working in lbortories, or those living in urbn or rurl communities were exmined for lctose (Lc) fermenttion nd ntibiotic resistnce. About two-thirds (62.7%) of the smples showed Lc' colonies only. Another 36.4% contined both Lc' nd Lc- orgnisms; however, mong these smples, the mjority of colonies (>9%) were Lc'. Six (<1%) hd Lc- orgnisms only. The Lc phenotype distribution ws not different between those individuls known to be tking ("on") or not tking ("off') ntibiotics. Since the Lc- popultions were commonly bsent or comprised minor popultion, we focused on Lc+ orgnisms s representtive of the generl phenotype of the totl fecl flor. We compred the ntibiotic resistnce profiles of the Lc' fecl flor found in smples from mles nd femles (Fig. 1). More thn 3% of the fecl smples from both groups hd >1% of orgnisms resistnt to mpicillin (Apr), streptomycin (Sm%, tetrcycline (Tcr), or cephlothin (Cf). Between 1 nd 25% of the smples hd -5% of the bcteri resistnt to these drugs. A qurter of the smples contined.1% bcteri resistnt to knmycin (Kn), nd 8 to 1% were predominntly resistnt (.5% level). There ws much less resistnce to chlormphenicol (Cm), gentmicin (Gmr), or nlidixic cid (Nr) (Fig. 1). No significnt differences were found between mles nd femles; therefore, the dt from these two popultions were combined for ll subsequent nlyses. No significnt differences were seen mong smples from different lbortory workers (dt not shown), so these smples were combined for nlysis. We exmined the individuls off ntibiotics in the popultion smpled. In mjority (62.5%) of ll fecl smples, 1% or more of the flor were resistnt to one or more of the ntibiotics tested. Over one-third (37.9%) of the smples showed resistnce(s) in.5% of the flor (Tble 2). When the people off ntibiotics were subdivided into hospitlized nd mbultory groups (Tble 2), it ws seen tht fecl smples from hospitl ptients hrbored more Apr strins t both frequency levels (P <.1). At the.5% frequency level, smples from the ptients showed more resistnce to knmycin, but not significntly (P <.1). There were, however, no other significnt differences in the smples from (n ' E (n FM FM FM FM FM FM FM FM Ap Sm Tc Cf Kn Cm Gm N FIG. 1. Antibiotic resistnce mong Lc' flor in fecl smples from femles (F) nd mles (M). The reltive mount of resistnt colonies (21 or 25% of Lc' bcteri on ech McConkey gr plte) in ech fecl smple ws recorded for ech ntibiotic (given on bottom line). Ap, Ampicillin; Sm, streptomycin; Tc, tetrcycline; Cf, cephlothin; Kn, knmycin; Cm, chlormphenicol; Gm, gentmicin; N, nlidixic cid. Dt for cephlothin cme from the hospitl group only.

VOL. 32, 1988 ANTIMICROBIAL RESISTANCE IN HUMAN FECAL FLORA 183 TABLE 2. Frequency of ntibiotic resistnce in individuls not receiving ntibiotics Popultion Resistnce % % Resistnt to given ntibioticc (smple size) frequency (%) Resistntb Ap Sm Tc Kn Cm Gm N Ptients (189) 21 67.2 48.2d 34.4 38.1 24.9 4.2 7.4.5 25 45. 27.e 21.2 22.8 7.9f 1.6.5.5 Nonptients (289) 21 59.5 34.9d 33.6 3.1 2.4 3.8 4.5.7.5 33.5 16.7e 15.6 18.7 3.5f 1.7.4.7 Totl (487).1 62.5 39.6 33.5 32.8 22.2 3.9 5.9.6 25 37.9 2.3 17.5 19.9 5.1 1.6.4.6 Of totl Lc+ orgnisms. b Smples with designted frequency of resistnce to ny of the seven drugs. Ap, Ampicillin; Sm, streptomycin; Tc, tetrcycline; Kn, knmycin; Cm, chlormphenicol; Gm, gentmicin; N, nlidixic cid. d x2 = 7.74; P <.1. e x2 = 6.87; P <.1. fx2 = 3.76; P <.1. g Includes nine hospitl stff members. the two groups off ntibiotics in terms of resistnce to the other five ntibiotics tested. A comprison mong the individuls off ntibiotics from the four subpopultions studied reveled tht one-fourth or more of ech group produced smples which contined Apr, Smr, nd Tcr (in the sme or different bcteri) t the.1% frequency level (Fig. 2A). Resistnce to knmycin ws somewht lower. In ll popultions, <1% of the smples crried Cmr, Gmr, nd Nr bcteri t the.1% frequency level (Fig. 2B). There were no significnt differences mong the three mbultory groups. The generl trend, however, ws towrds lower numbers of resistnt bcteri in the rurl smples, nd Cmr ws bsent from this popultion. In contrst to other groups, pproximtely 4% of rurl smples hd >1% Nr bcteri (Fig. 2B). This finding ws due to n Nr Escherichi coli isolte crried by severl members of single fmily. A 6 E '1% ;.5% 2. E H L U R H L U R H L U R H L U R Ap Tc Sm Kn 2qr 1F B UunH n n H L U R H L U R H L U R Gm N Cm FIG. 2. Frequency of ntibiotic-resistnt Lc' orgnisms in fecl smples from four humn popultions not ingesting ntibiotics: hospitl (H), lbortory (L), urbn (U), nd rurl (R). See legend to Fig. 1. Significnt differences in mpicillin resistnce were found mong hospitl, urbn, nd rurl groups (see text). Severl differences ppered when the smples from ech mbultory group were compred with those from the hospitlized ptients (Fig. 2A). Both the urbn nd the rurl group smples hd significntly less Apr thn the hospitl group (urbn: x2 = 4.36, t the 1% level only, with P <.5; rurl x2 = 6.17 t the.1% level, with P <.5, nd X2 = 9. t the.5% level, with P <.1). Effect of ntibiotic usge on frequency of resistnt orgnisms. Most of the ntibiotic usge mong people tested in this study occurred in the hospitl. Fifty-three ptients (of the 247 ptients whose ntibiotic use ws known) were receiving one or more ntibiotics. An dditionl 12 of 314 mbultory donors (whose ntibiotic use ws known) hd tken n ntibiotic within 2 weeks prior to smpling. Given the smll number of nonhospitlized individuls on ntibiotics nd the miniml difference found mong hospitlized nd nonhospitlized individuls off ntibiotics (Tble 2), we focused on the hospitlized group to determine the effect of ntibiotic use. The only significnt difference ws the higher percentge of fecl smples with Apr bcteri from the group on ntibiotics (t the.1% frequency level, x2 = 6.7, P <.1; t the.5% resistnce frequency, x2 = 11.9, P <.1). Detectble, but not significnt, differences were lso noted to four other ntibiotics: streptomycin, tetrcycline, knmycin, nd cephlothin (Fig. 3). Frequency of more thn one ntibiotic resistnce determinnt in fecl flor. Fecl smples commonly contined more thn one resistnce determinnt in the sme or different orgnisms. Almost 4% of ll smples hd multiple determinnts t.1% levels (Fig. 4). Slightly greter numbers of smples from those on ntibiotics contined more thn one resistnce determinnt (two or more drugs) t the.1% frequency level compred with smples from those not tking drugs (Fig. 4). This difference ws significnt when - more thn four resistnces were present (23 versus 1%; P.1). At the higher frequency level (-5%), significnt differences were found between the two groups in the number of smples with multiple resistnce determinnts (Fig. 4). The group on ntibiotics hd two- to threefold more smples with more thn two nd three resistnces nd ninefold more smples with more thn four resistnces. A totl of 364 smples were rndomly chosen from mong those off ntibiotics for nlysis of multiple resistnce within the sme orgnism (coresistnce). Seventy-one percent crried detectble levels of one or more of 31 different coresistnce ptterns. The most frequently occurring ptterns were

184 LEVY ET AL. W, ; E 4-2 3- I- 1 %.5% ' off on off on off on off on off on off on off on off on Ap Sm Tc Cf Kn Gm Cm N FIG. 3. Effect of ntibiotic usge on ntibiotic resistnce frequency. Fecl smples from ptients off nd on ntibiotics were compred for frequency of resistnt bcteri t two levels. See legend to Fig. 1. Significnt differences ppered in mpicillin resistnce t both levels (see text). Smr Tcr (22.3%), Apr Smr Tcr (16.5%), nd Apr Smr Tcr Knr (14.6%). Twelve other coresistnce ptterns were detectble in.1% of positive smples (Tble 3). Comprison of resistnce between single- nd multiplesmple nlysis. We compred the dt generted from single smples with those generted from ll smples. No sttisticlly significnt differences were found t ny resistnce level to ny of the ntibiotics (Tble 4). Further exmintion of dt from the multiple-smple group reveled tht chnges in resistnce ptterns of gut flor were extremely common nd occurred frequently within 2-week period. Some 143 pirs of resistnce ptterns were exmined, using the first two smples obtined from ech individul of the mbultory group off ntibiotics. Ninety percent of ll individuls showed gin (47.6%) ndlor loss (65.7%) of one or more detectble resistnces. Forty-four percent exhibited gin or complete loss of resistnce mrker t the.1% level. cl C') 5- >3 4 >4 3-4- ntibiotic sttus off on off on resistonce frequency > 1% > 5 % FIG. 4. Frequency of different resistnce determinnts (more thn two, three, or four) in fecl smples from individuls on or off ntibiotics. Significnt vlues were s follows:, x2 = 8.3, P <.1; b, x2 = 8.1, P <.1; c, x2 = 2.4, P <.1; d, x2 = 26.6, P <.1. TABLE 3. ANTIMICROB. AGENTS CHEMOTHER. Frequency of common coresistnce ptterns Coresistnce No. of Frequency Coresistnce ptternsmples' (%)C Smr Tcr 58 22.3 Apr Smr Tcr 43 16.5 Apr Smr Tcr Knr 38 14.6 Smr Tcr Knr 17 6.5 Apr Tcr 19 7.3 Apr Smr 19 7.3 Apr Knr 13 5. Smr Knr 13 5. Apr Smr Knr 12 4.6 Tcr Knr 8 3.1 Apr Smr Tcr Knr Gmr 7 2.7 Apr Smr Tcr Knr Cmr 5 1.9 Knr Gmr 5 1.9 Tcr Cmr 5 1.9 Smr Cmr 4 1.5 Only those coresistnce ptterns occurring in 1% or more of the smples re presented. b Number of fecl smples of 364 tested for coresistnce. c Frequency of coresistnce pttern mong 26 smples which contined coresistnce. Repet study of the urbn group. The urbn-dwelling group of medicl students smpled in the 1987 survey consisted only of those off ntibiotics. In this group, 46.3% of smples showed 1% or more of the flor hving Apr, Smr, Tcr, Knr, or Tmr strins. Resistnce(s) in 25% of the flor ws found in 26.4% of the smples. Resistnce to the individul drugs t the -1% level ws s follows: Apr, 25%; Smr, 28.8%; Tcr, 38.8%; nd Knr, 15%. High-frequency resistnce (.5% of the fecl flor) ws 11.3% for mpicillin, 17.5% for streptomycin, 21.3% for tetrcycline, nd 7.5% for knmycin. Trimethoprim resistnce (not ssyed in the erlier study) ws found in 7.5% of the smples from this group. Only few smples (2.5%) bore this resistnce in.5% of the flor. These findings were very similr to those of the urbn group studied 8 to 9 yers erlier. Multiple resistnce in smples from the medicl students in the 1987 survey ws very similr to tht obtined from the off ntibiotic group of the erlier study. Thirty-one percent of the smples contined multiple (more thn two) resistnce determinnts t the.1% level, 17.5% bore three or more resistnces, nd 1% hd flor with resistnce to ll four drugs tested in both studies. At the.5% level, 15% hd multiple (more thn two) resistnces, 7.5% hd more thn three, nd 6% hd four. DISCUSSION Considerble ttention hs been given to ntibiotic resistnce found mong pthogenic strins of bcteri cusing disese in humns, other nimls, nd plnts. The present study determined the reltive frequency of potentil reservoirs of these resistnce genes residing in the fecl flor of humns t lrge. We exmined mbultory donors s well s hospitlized ptients with nd without history of hving tken ntibiotics in the previous 2 weeks. This time period ws selected becuse previous studies hve shown tht chnges in fecl flor subjected to ntibiotics were usully reversed by 1 to 14 dys fter stopping the drug (8, 14). In second, more recent survey of urbn dwellers, ll individuls reported tht they hd not tken n ntibiotic for 6 months. The findings in the two studies were similr. More thn 6% of the smples from those not tking n ntibiotic showed resistnce to t lest one drug t level of

VOL. 32, 1988 ANTIMICROBIAL RESISTANCE IN HUMAN FECAL FLORA 185 TABLE 4. Resistnce of Lc' flor in single- versus multiple-smple nlyses Anlysis nd Frequency Resistnce (%)' smple size level (%) Ap Sm Tc Kn Cm Gm N Single smple 21 42.2 36. 34.4 22.3 4.3 5.6.9 636b 25 24. 2.3 22.3 7.2 2.2.5.5 Multiple smple.1 41. 35.7 32.6 25.5 4. 7.4.9 1,43.5 23.4 2.4 2.4 8.3 2. 1.1.5 For definitions of drug bbrevitions, see footnote c to Tble 2. b Excludes four with no Lc+ orgnisms..1% of the totl lctose-fermenting orgnisms (Tble 2). Almost 4% of ll smples from this group contined two or more different resistnce determinnts t this level (Fig. 4). Of these, much of the multiple resistnce occurred in the sme orgnism (Tble 3). Common resistnces were Apr, Smr, nd Tcr, followed closely by Knr. Cf' ws exmined in the hospitl popultion nd ws lso found to be high. Cmr, Gmr, or Nr ppered in <1% of the smples. Using vriety of methods, other investigtors in different countries hve exmined the frequency of resistnce in fecl flor of smller smple sizes of 25 to 1 donors. These studies were performed in the lte 196s nd erly 197s primrily with hospitl-ssocited personnel nd ptients with no known ntibiotic ingestion (3, 15, 16, 18). While it is difficult to compre these studies due to differences in ntibiotic selection nd methodology, in generl, they showed tht bout 2 to 4% of smples bore resistnce to one or more drugs in 1% or more of the fecl flor (3, 15, 16, 18) Ȯnly smll number of studies hve exmined helthy, mbultory popultions (11, 13, 17). The most extensive of these, performed in 1968 to 197, exmined 39 children nd dults from urbn nd rurl res of Englnd (11). Using both quntittive nd qulittive methods, this group found tht ll smples contined some level (usully low) of orgnisms resistnt to mpicillin, streptomycin, tetrcycline, knmycin, chlormphenicol, nlidixic cid, nitrofurntoin, or sulffurzole. Only 17% of smples hd resistnce in isoltes selected from the most common morphologic phenotype. We found n even higher prevlence of resistnce genes in the norml gut bcteri of people in the Boston re thn ws reported from these previous, more limited studies. Since we did not exmine sulfonmide resistnce, which in other studies occurred s frequently s tetrcycline (18), our estimte of resistnce frequency is probbly lower thn if this drug hd lso been tested. The low frequency of Gmr, Nr, nd Cmr my reflect lower generl usge of these drugs, but this does not explin the high frequency of Smr in the fce of miniml usge of this drug tody (9) nd in the popultion we studied (dt not shown). Previous reports hve documented incresed infection or gut coloniztion with resistnt bcteri during hospitliztion, with or without ntibiotic tretment (3, 12, 15, 16, 19). One study observed n lmost twofold increse in incidence of Cmr nd Tcr strins in ptients not tking ntibiotics who were hospitlized for 2 weeks (19). The resistnce frequency dropped fter leving the hospitl. In the present study we found only miniml difference in resistnce to the individul drugs in fecl smples from the hospitlized individuls off ntibiotics s compred with mbultory groups (Tble 2). Even when the presence of two or more resistnce determinnts in the smple ws exmined, hospitlized nd mbultory groups off ntibiotics were not sttisticlly different. A history of recent ntibiotic tretment mong hospitlized ptients correlted with significnt increse only in the frequency of Apr nd not resistnce to the other drugs (Fig. 3). Although we lso noted tht Apr ws the only significnt difference between noningesting hospitlized nd mbultory groups (Fig. 2), we believe this increse is linked to ntibiotic ingestion since our comprison delt with only the hospitlized group. Therefore, both hospitliztion nd ntibiotic ingestion re involved in sttisticlly significnt increse in Apr. We lso observed significnt increse in the reltive numbers of different ntibiotic resistnce determinnts in ech fecl smple of ntibiotic ingestors (Fig. 4). Still, this ntibiotic effect ws less evident thn the two to threefold chnges reported in previous studies (3, 15, 16, 19). This my be relted to n lredy higher bse-line level of resistnce in the popultion studied here. Antibiotic usge did not led to chnge in the lctose fermenttion profile of the gut flor, demonstrting tht resistnce determinnts in E. coli nd other lctose-fermenting gut bcteri were sufficiently common tht chnge in reltive distribution of the orgnisms did not occur. Although most individuls showed gin or loss of resistnt bcteri when multiple smples were nlyzed, there ws no effect on the popultion s whole (Tble 4). The constnt chnging of the fecl E. coli popultion hs been reported by others (2). Our findings suggest tht these bcteri represent lrge common pool of resistnce determinnts which re mintined t prticulr levels in people even from different fmily, residentil, nd occuptionl groups, while they my chnge within different fecl smples from the sme individul. To our knowledge, this is the first such extensive comprtive evlution on the frequency of ntibiotic resistnce genes in the gut flor of mbultory nd hospitlized people. Although the initil study ws performed in the lte 197s, the recent smpling of members of one of the sme popultions (medicl students) not ingesting ntibiotics showed similr frequencies of both single nd multiple resistnce. Consequently, we believe tht our erlier findings reflect the present-dy frequency of ntibiotic resistnce in the fecl flor of n mbultory popultion, in which the recent study showed tht 7.5% of the smples hd significnt (.1%) levels of trimethoprim resistnce. A mjor portion of the erobic grm-negtive gut flor (t lest in the Boston re) now consists of resistnt bcteri. About 16 to 18 E. coli re generlly found per grm of humn feces (1). Given bout 1 g of feces excreted per dy, n individul with 1% frequency of resistnce determinnt produces 17 to 19 resistnt bcteri per dy. Even t 1% resistnce level, reltively lrge mounts of resistnce determinnts re being excreted. This represents

186 LEVY ET AL. lrge reservoir of resistnt bcteri nd resistnce genes. This sitution my be the consequence of slow loss (i.e., more thn 6 months) of resistnce determinnts cquired from previous ntibiotic selection or by ingestion of resistnt strins in foods (7) or by other unidentified fctors which fvor their persistence (6). Whtever the cuse, these findirigs show the present-dy crrige of high numbers of resistnt bcteri nd resistnce genes in the humn gut flor in the bsence of concurrent or recent ntibiotic consumption. ACKNOWLEDGMENTS This study ws supported in prt by Public Helth Service contrct NOIAI72529 with the Ntionl Institutes of Helth nd by grnt from the Environmentl Protection Agency under ssistnce greement CR813481-1-1 to the Center for Environmentl Mngement of Tufts University. We pprecite the invluble help of George Stulker in processing nd retrieving the dt for these nlyses. We thnk Toni Keller, Hollington Lee, Kte Mel, Stnley Miller III, nd Dine Petrowski for ssistnce in microbiologic testing. We thnk Jon Downing, Brry Golden, Anne Mcone, Mrin Richter, Anne Summers, nd Kren Tlmdge for helping-to obtin smples. LITERATURE CITED 1. Buer, A. W., W. -M. M. Kirby, J. C. Sherris, nd M. Turck. 1966. Antibiotic susceptibility testing by stndrdized single disk method. Am. J. Clin. Pthol. 45:493-496. 2. Cooke, E. M., S. Edwins, nd R. A. Shooter. 1969. Chnging fecl popultion of Escherichi coli in hospitl medicl ptients. Br. Med. J. 4:593-595. 3. Dtt, N. 1969. Drug resistnce nd R fctors in the bowel bcteri of London ptients before nd fter dmission to hospitl. Br. Med. J. 2:47-411. 4. Frrr, W. E. 1985. Antibiotic resistnce in developing countries. J. Infect. Dis. 152:113-116. 5. Lederberg, J., nd E. M. Lederberg. 1952. Replic-plting nd indirect selection of bcteril mutnts. J. Bcteriol. 63:399-46. 6. Levy, S. B. 1982. Microbil resistnce to ntibiotics: n evolving nd persistent problem. Lncet i:83-88. 7. Levy, S. B. 1984. Antibiotic resistnt bcteri in food of mn nd nimls, p. 525-531. In M. Woodbine (ed.), Antimicrobils nd griculture. Buttersworth, London. ANTIMICROB. AGENTS CHEMOTHER. 8. Levy, S. B. 1986. Ecology of ntibiotic resistnce determinnts, p. 17-29. In S. B. Levy nd R. P. Novick (ed.), Antibiotic resistnce genes: ecology, trnsfer nd expression. Cold Spring Hrbor Lbortory, Cold Spring Hrbor, N.Y. 9. Levy, S. B., J. P. Burke, nd C. K. Wllce (ed.). 1987. Antibiotic use nd ntibiotic resistnce worldwide. Rev. Infect. Dis. 9(Suppl.):231-316. 1. Levy, S. B., B. Mrshll, A. Onderdonk, nd D. Rowse-Egle. 198. Survivl of E. coli host-vector systems in the mmmlin intestine. Science 29:391-394. 11. Linton, K. B., P. A. Lee, M. H. Richmond, W. A. Gillespie, A. J. Rowlnd, nd V. N. Bker. 1972. Antibiotic resistnce nd trnsmissible R fctors in the intestinl coliform flor of helthy dults nd children in n urbn nd rurl community. J. Hyg. 7:99-14. 12. McGown, J. E., Jr. 1983. Antimicrobil resistnce in hospitl orgnisms nd its reltion to ntibiotic use. Rev. Infect. Dis. 5: 133-148. 13. Moorhouse, E. C. 1969. Trnsferble drug resistnce in Enterobcteri isolted from urbn infnts. Br. Med. J. 2:45-47. 14. Richmond, M. H. 1977. The survivl of R plsmids in the bsence of ntibiotic selective pressure, p. 61-7. In J. Drews nd G. Hogenuer (ed.), Topics in infectious diseses, vol. 2. R. fctors: their properties nd possible control. Springer-Verlg, Berlin. 15. Slzmn, T. C., nd L. Klemm. 1967. Trnsferble drug resistnce (R fctors) in Enterobctericee: reltionship to nosocomil infections, p. 212-22. Antimicrob. Agents'Chemother. 1966. 16. Shw, E. J., N. Dtt, G. Jones, F. M. Mrr, nd W. J. B. Frond. 1973. Effect of sty in hospitl nd orl chemotherpy on the ntibiotic sensitivity of bowel coliforms. J. Hyg. 71:529-534. 17. Smith, H. W., nd S. Hlls. 1966. Observtions on infective drug resistnce in Britin. Br. Med. J. 1:266-269. 18. Sogrd, H. 1975. Incidence of ntibiotic resistnce nd trnsmissible R fctors in the grm-negtive bowel flor of hospitl ptients on dmission. Scnd. J. Infect. Dis. 7:253-258. 19. Somnpolinsky, D., V. Yron, nd W. J. Alkn.'1967. Microbiologicl chnges in the humn fecl flor following the dministrtion of tetrcyclines nd chlormphenicol. Am. J. Proctol. 18:471-478. 2. World Helth Orgniztion Scientific Committee on Antimicrobil Resistnce. 1983. Antimicrobil resistnce. Bull. W.H.O. 61:383-394, 423-433.