Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
174 plant , 4 7 % of poultry were contaminated with Campylobacter iejuni and 4 . 7 % were contaminated with Salmonella spec ies . l 3 Contamination with Campylobacter spp . was found in 2 2 % o f poultry from reta il sources and 3 . 5% with Salmonella spp . Contamination with ï¿½ jeiuni was observed in 0 . 4 % o f beef samples , but no salmonella contamination was found . Salmone l l a contamination was 2 . 7 % in pork products . In this study , 30% o f the salmone l l a isolates from reta i l poultry were res istant to tetracycl ine . These findings show that ï¿½ QQli , Salmonella spp . , and ï¿½ - iejuni commonly are found on meat and poultry products . Human ingestion of these bacteria might result from contamination of hands during food preparation or consumpt ion of inadequately cooked animal food products . Colonization of the human intestine by ant ib iot icÂ res istant ï¿½ Â£Qli, in the absence of ant ib iot ic use , has occurred fol lowing handl ing of commerc ial ly prepared chicken carcasses in the kitchen . The abil ity of part icular ï¿½ Â£Qli stra ins to colon i z e the intest ina l tract both of humans and various spec ies of anima l s depends on the presence of colon i z at ion factors and spec i f ic cel l surface characterist ics common to both , because many of the 0-serotypes of ï¿½ Â£2li found in w oultry , p igs , and calves also have been found in humans . 7 , There fore , it is l ikely these ï¿½ Â£Qli are from a common pool . It has been shown that ï¿½ Â£211 stra ins from the al imentary tract of humans and chickens are ident ical by O , H , K serotyping , by antimicrobial res istance patterns , and by plasmid restriction endonuclease profi les . 7 , 8 Also , serotype ident ity among E . coli stra ins of one speci f ic serotype ( 0 2 : K l ) have been identi f ied commonly in human urinary tract infect ions and neonatal meningitis and in animal d isease ( bovine mastitis , and chicken septicemia ) . Recently , a group of such stra ins of both human and animal origin was submitted to clonal analys is by compari son of outer membrane prote in ( OMP ) patterns , l ipopolysacehharide patterns , electrophoretic mob i l ities of enzymes , b iotyping , etc . l Human isolates were found to fit into two clonal groups , poultry isolates belonging to one and bovine isolates to the other . Human isolates of one clonal group were d ist inguishable from poultry isolates of the same group by the ir plasmid content : human isol ates of the second c l onal group were dist inguishable from bovine isolates of that group by a minor alterat ion in the OMP pattern and by the ir plasmid pattern . Whether these dif ferences in plasmid pattern ( or in the OMP pattern , in the case of bovine isolates ) indicate that the populations of human and animal isolates are not overlapping , even though very s imilar , is unclear . I n view of the exchange of ï¿½ QQli and Salmonella spp . that can occur between food animals and humans , movement of ant imicrobial -re s istance genes from the intest inal flora o f anima ls t o the flora of humans may occur b y carriage of
175 plasmids and transposons . Such movement of ant imicrob ial Â res istance genes may fol low movement t o and pers istence in the human a l imentary tract of the foodborne enteric bacteria or from subsequent conj ugative trans fer of the plasmid to a res ident constituent of the human intestinal flora . Conj ugat ive trans fer of R plasmids can be detected in the human intestinal tract in the presence of an ant imicrob ial that a l l ows an increase in the number of R+ donor cel l s and other cel l s that have received the R plasmid . 2 However , such trans fer might not occur commonly in the absence of antimicrob ial selection 1 2 in humans , although it occurs qu ite e fficientl y in the rumen of sheep a fter 24 hours of starvation . 14 The foregoing suggests that the populations of enteric bacteria of animal and human origin might be cons idered as a common pool of ant imicrobial -res istance genes ( transposons , R plasmids , and chromosomal genes ) capable of being ampl i f ied through ant imicrobial exposure and subsequent selection . DIS SEHIBAT ION OF BESI STAHCE GENES AND GENOKES Use of each new antimicrobial agent introduced over the past hal f-century has caused the emergence and global dissemination of bacterial genes encoding res istance to the agent . Growing prevalence of genes that encode res istance to older agents has prompted development and use of new ones , which have caused succeeding rounds of emergence and spread of new res istance genes . Disseminat ion of a res istance gene incurs d i fferent kinds of costs as it proceeds . When the res istance is not recogni z ed or when opt imal medical ski l l s , laboratory services , or newer antimicrobial agents are not promptly ava ilable , the health burden is treatment failure with prolonged morbidity or death . When optimal support is ava i l able , which is rare everywhere at first and seldom in poorer regions , costs shi ft towards the expense of the support and the costly new agents and the toxic ity of some of the agents . Recognition of the emergence and spread of res istance and of its costs initial ly ra ised fears that the activities encoded by the emerging and spreading res istance genes would exceed our abi l ity to develop new agents . However , nearly al l of the target s ites in bacteria that are exploited by existing agents were explo ited in the first-quarter century of the antimicrob ial era . The f inding of few new target s ites in the second-quarter century suggested that these sites were an unreplenishable resource--one increas ingly endangered by pro l i ferat ing resistance genes that prevented intact ant imicrobial molecules from reaching the target sites .
17 6 The fear that we would run out of e ffective ant imicrobial agents altogether was greatly diminished in the past decade by the introduction of many new agents that evaded the e ffects of existing res istance gene products and reached and inhib ited the old target s ites . Within the past year , however , a number of new res istance genes have been detected in different parts of the world that inactivate many of the largest class of the newer agents , the thirdÂ generation cephalosporins . It was recogni z ed early that use of antimicrobial agents was the maj or force driving the emergence and dissemination of res istance genes , and that such use should therefore be reduced to its essential minimum . What was not c l ear earl ier , however , was the interrelatedness of what might be cal l ed a global system of antimicrobial res istance and the consequent effect of use in one area upon res istance in another . New evidence for this has been developing from both molecular and epidemiological work . The molecular studies show that res istance genes and the plasmids that carry them constitute intricate assembl ages of multifunctional modular components with the size and complexity but not the packaging of viruses . For such a genome to have arisen de novo in a patient or his neighbors in an intens ive care unit would be the equivalent of spontaneous generation . Each must have had a l engthy evolut ionary history . Studies in molecular biology and molecular genetics are beginning to suggest some of that evolutionary history . Individual res istance genes evolve from ancestral genes , are moved to other genomes or transposons or by s ite-spec i f ic recombination , acquire promoters , become l inked to genes under dif ferent selection , are trans ferred on conj ugative plasmids to other strains occupying other niches in bacterial ecosystems , and are carried in bacteria to other bacterial habitats . Each such event in the evolut ion of a res istance gene or its plasmids may initiate a new stage in its dissemination by extending its range or pers istence . And the chance of occurrence of each such event would be greatly enhanced by antimicrobial use , which ampl i fies at every step the prevalence of the gene and its genome and hence the chance that someth ing new wi l l happen t o them . Bes ides showing that res istance genes and the ir plasmids must have extended l ineages , the molecular work is now a l so beginning to trace some of those l ineages . Genetic maps of the large transposon Tn 2 1 suggest that it carried mercury res istance as wel l as res i stance to several of the early ant imicrobials be fore be ing included in the first recogn i z ed res istance plasmids ( in shigel los i s in Japan in the late fi fties ) . This transposon has subsequently turned up in plasmids in many parts of the world , including most of the varied plasmids that first brought gentamicin
177 res istance t o German medical centers and plasmids carrying a variety of d i f ferent P-lactamase genes . A coro l l ary of the extended l ineages of res i stance genes and plasmids is that the res istance observed in the bacterial populations of one patient or one medical center is a consequence of prior use of antimicrobials , not j ust there but in other bacterial populat ions as wel l , including others that may have been remote in time and place . Th is growing understanding of the interrel atedness o f the res istance observed in the world ' s interconnect ing bacterial populations intens i f ies earl ier concerns about ant imicrobial agents as animal feed additives . Anima l s get nearly ha l f of the antimicrobials used in the United States , and the pool of res istance genes and genomes in the ir flora may be much greater than that in humans . Bacteria of animal origin are not a remote and separate populat ion , moreover , but enter most households cont inual ly on slaughtered animal carcasses . I f use of antimicrobials in one bacterial populat ion a ffects prevalence of res istance in other bacterial populat ions more than sl ightly , then antimicrobial additives in animal feeds would contribute s ign i f icantly to res istance in human flora . Epidemiologic observations in the l ast few years has added to these examples of spec i f ic res istance plasmids that are found in isolates of bacteria from both anima l s and humans in the United states , and has in some cases reconstructed a path of spread from animal to human . These examples have thus far been observed in isolates o f Salmonel l a spp . , thus their elaborate serotyping by a network of medical and veterinary laboratories made them pecul iarly traceabl e and particularly suited at this stage to risk assessment model ing . It should be empha s i z ed , however , that salmonel lae are a sma l l part of the aerobic flora in the gastrointestinal tract of animals , and an even sma l ler part of that of humans ; and these sa lmonel lae represent less than 1% of the bacteria aga inst which therapy in humans i s directed and in which res istance may b e a problem . The concerns outl ined above , although now best exempl i fied by salmonel l ae , apply to a l l bacteria that infect humans . PAUCITY OF DIBECT EVIDENCE There is l ittle evidence d irectly l inking subtherapeutic use o f penicil l in and tetracycl ines in animal feeds to human infect ions with pathogen ic bacteria . As summa ri z ed in Chapter V , there is good evidence from only two studies that non-salmonel l a enteric organisms in which ant imicrob ial res i stance was induced by the subtherapeutic administrat ion of ant imicrobial agents might be spread from anima l s to humans . 5 , 6 Two other studies 4 , 10 fa i l ed to show that
178 multiple-drug-resistant stra ins of animal origin cause infect ion in humans exposed to these stra ins , but it is poss ible that the power of these studies to detect such infections was too low to make a negative result meaning ful . A number of studies have attempted to demonstrate that outbreaks of drug-resi stant salmonel los i s in humans may be attributed to the administration of antimicrob ial agents in subtherapeutic dosages to farm animals : however , in a l l but one instance there has been some de fect in the proof of the cha in of transmiss ion ( see Chapter V ) . A convinc ing case was an outbreak of infection due to chloramphenicol -res istant ï¿½ newport in which infect ions in humans were attributed to the ingest ion of ground bee f from animals medicated with chl oramphenicol ( a drug not approved by regulation for use in bee f animal s ) : 1 5 evidence for identity of the stra ins through the cha in of transmiss ion was based both on the unusual pattern of ant imicrobial res istance and on plasmid ana lys is . In summary , whereas the theoretical bas i s for concern that the subtherapeut ic admini strat ion of ant imicrobial agents to animals may foster infect ions by drug-re s istant pathogens in humans is immense , the direct evidence of such an ef fect is sparse and general ly indirect . There are many pos s ible reasons why such an ef fect might be d i f f icult to detect : o The degree to which ant imicrob ial agents are administered for subtherapeut ic purposes is generally unknown , varies from farm to farm and from t ime to t ime , and is not generally monitored : l ikewise , the proport ion of drugs given for subtherapeut ic , as opposed to therapeut ic , purposes i s not wel l de f ined . o The relat ive influence of subtherapeut ic use ( growthÂ promot ional and prophylact ic administration ) and therapeut ic use on the preva lence of drug-resi stant stra ins is not known with certainty and may differ among drugs . o The prevalence of salmonel lae , both suscept ible and res i stant , in various food products is not rout inely monitored , except in spec ial circumstances , such as outbreaks : even in special circumstances , it is genera l l y assessed only retrospect ively , when the s ituation may be qu ite d i f ferent from that at the start of the outbreak (when samples of suspect food products are not ava ilable ) . o Most cases of salmone l l os i s are unident i f ied , and in only rare outbreaks is an ef fort made to ident i fy the source of the infect ing organism : no ef fort to ident i fy a source i s made in sporadic cases .
179 o There are many sporadic cases of salmonel los is which create considerable " background noise" for the invest igator attempting to def ine the cha in of transmiss ion of a spec i f i c salmonella stra in ; prec ise , rapid , and efficient molecular techniques for the ident i fication o f unique isolates ( i . e . , to prove the clonal identity of isolates ) have only recently become available . There i s no direct ev idence that establ ishes the proport ion of human multiple-drug-res istant salmonel lae that is of animal origin or the proportion due to person-to-person transmission . Only a sma l l proportion of multiple-drugÂ res istant salmonel lae in humans occurs as part of a recogn i z ed outbreak or epidemic . When such outbreaks have been detected and investigated , CDC surve i l l ance data have indicated that foods of animal origin are impl icated in almost 7 0% . 3 It i s reasonable to speculate that sporadic cases of human salmonel los is caused by multiple drug-res istant salmonel lae may occur as part of undetected outbreaks , and sti l l others are undoubtedly of indirect anima l origin , resulting from person-to-person spread . There is not an extens ive body of data bearing on this issue . Some invest igators bel ieve that the number o f cases of person-toÂ person spread of salmone l l os is has been underestimated , and that contamination of food products by human carriers , as wel l as animal sources , of multiple-drug-res istant salmonellae must be cons idered in the estimates of cases . Some mul t iple-drug-res istant salmonel lae , notably Salmonel l a we in , ( uncommon in the u . s . ) have n o apparent anima l source . Nevertheless , there is a general para l lel ism between the prevalence of multiple-drug-resistant sa lmonellae in anima l s and i n humans , and numerous investigators in the United states bel ieve , there fore , that the maj ority of human multiple-drug-res istant salmonel lae are , directly or indi rectly , ultimately o f animal origin . I t must be emphas i z ed again , however , that food processing techniques are des igned to prevent contaminat ion and transmiss ion of animal pathogens to humans via the food cha in . In the maj or ity of outbreaks of multiple-drugÂ res istant salmone l l ae in humans , it has been poss ible to demonstrate flaws or de fects in food process ing techniques that a l l owed the contaminat ion with salmonel lae . Such defects have nothing to do with whether the salmonellae are ful ly susceptible to ant ibiotics or are multiple-drugÂ res istant . Ultimately , therefore , the farm animal -to-human cha in of transmiss ion , of all sa lmone l lae w i l l be interrupted more rel iably by careful attent ion to accepted techniques of food processing and preparat ion than by any other publ ic measure that could be contemplated .
180 EFFECT OF DRUG BES I STAHCE ON HUMANS The committee has been asked whether drug res istance o f salmonellae caused by subtherapeutic admini stration o f antib iotics i n feed causes a n increase i n the number o f cases o f salmonel losis in humans or compl icates treatment of these cases . These questions are di fficult to answer , a lthough they are obviously fundamental to the assessment of risk . Drug-resistant Salmonella spp . infectious to both humans and animals could cause an increase in morbidity in humans in four ways : o By increas ing the overal l prevalence of these pathogens ( both res istant and suscept ible stra ins ) in anima l s or the ir food products , could increase the potential for exposure of humans to salmonel los is . Whether or not this increase occurs is unknown . The prevalence of res istant strains in animals might be increased by the subtherapeut ic administration of antimicrobial agents and the subsequent suppress ion of the normal gut flora of the anima l s ; that would be analogous to the " etiologic fraction" in humans . However , the prevalence of susceptible stra ins might be reduced concomitantly , with an overal l effect that i s d i f f icult t o predict . o By increasing the virulence of drug-res istant pathogens ( see Chapter I I I ) . It is unc lear whether virulence is increased ; some evidence suggests that virulence may be increased , other evidence , pos s ibly less convinc ing , suggests that the oppos ite result may occur . To the extent that the epidemiologic behavior of other res istant species is a guide to the ef fect of the widespread use of antimicrobial agents on the prevalence of pathogens in the environment , it cannot be determined that such resistance wi l l decrease the prevalence or virulence of the res istant species . overa l l , the inc idence of reported cases of salmone l l os i s in the United States has continued to rise , slowly but steadily , over recent decades concomitantly with evidence o f increas ing prevalence of drug res istance in the isol ates . Whether or not the increase in reported cases of salmonel losis is related to the subtherapeut ic use of ant ib iot ics in animal feed is not clear , of course ; but certa inly it cannot be said that the inc idence of this infection in humans has been decreas ing while the subtherapeutic administrat ion of ant imicrobials to animal s has been steadi ly increas ing . However , many other confounding factors make it d i f f icult to determine the cause-and-e ffect relation between the subtherapeutic administrat ion of ant imicrobials to anima l s and the increas ing number of cases of salmonel losis caused b y both suscept ibl e and res istant isolates . Among these
18 1 confounding factors is the increasing prevalence o f " fast food" in the American diet : these foods are prepared o ften in large batches wherein a small amount of contaminating bacteria may have a magn i fied impact . o By evoking the effect o f the " etiologic fraction . " Evidence clearly indicates that in some individual s infected with drug-resistant stra ins of salmonel lae the infection is susta ined because , prior to infection , they were ingesting antimicrobial agents to which the bacterial stra ins were res istant . It is bel ieved that these individual s would not have been infected had the stra ins been suscept ible . o By interfering with the efficacy of antimicrobial treatment . There are some patients infected by drugÂ res istant stra ins of nontyphoidal salmonel lae for whom ant imicrobial treatment is ineffective because the pathogens are res istant . The committee bel ieves that such instances are rare at the present time . SUBTBEBAPEQTIC VS . THERAPEUTI C POSES The presence of ant imicrobial agents in the environment obviously causes the selection of microorganisms that are res istant to those agents . The clearest example of this phenomenon is seen in the in vitro determinations of antimicrobial res istance of isolates of bacteria , which are performed da ily in cl inical microbiology laboratories . With the rare exception of chromosomal ly mediated drug res istance , most such res istance is due to trans ferable res istance factors , or R plasmids . By def inition , the minimal inhibitory concentration ( MIC) of an antimicrobial agent for a given bacterium is the lowest concentration that inh ib its completely the growth of the organism . At sub-MIC concentrations there can sti l l be measurable , dose-dependent growth inhibition that is not complete . In determining the relat ive ef fects on drug res istance of subtherapeutic vs . therapeutic concentrations of antimicrobial agents , several cons iderations must be evaluated : ( 1 ) how often and for how long does the concentration of the drug reach or exceed the MIC? ( 2 ) how quickly do the res istant organisms grow during this period? ( That is related to evaluat ing the relat ive growth advantage of the res istant vs . the susceptible organisms . ) ( 3 ) at concentrat ions of drugs below the MIC , is there a doseÂ related ef fect on the e f f iciency of R-plasmid trans fer? Virtua l ly a l l work to date on drug res istance involves the study of clona l ly pure s ingle stra ins of bacteria . Thus , the important issue of spread of drug res istance via R plasmids from res i stant to susceptible stra ins , part icul arly
182 of d i fferent spec ies , has not been ful ly explored a t a range of drug concentrations that would shed l ight on the d i fferential e f f iciency of drug-resistance selection . It is nevertheless possible to speculate on the ef fects . Assume the s imple case of two stra ins in the env ironment at equal inocula ; one stra in possesses a transferable R plasmid and i s drug-resistant , the other stra in lacks an R plasmid and is drug-susceptible , but can acquire drug res istance on acqu i s ition of the R-plasmid by conj ugative trans fer . Cons ider the effects on the environment of several d i fferent concentrations of drugs . At high , super-MIC , drug level s , only the res istant stra in surv ives . There i s a net increase in drug res istance , as a consequence of continued growth o f the res istant stra in , but there is n o spread of res istance . Al l res istance increase is from clonal expans ion . At low enough drug levels ( i . e . , sub-MIC ) there is no select ive ef fect of the ant ibiotic . At MIC ( bacteriostat ic , but not bacteric idal ) , there is selection and expans ion not only of the original R-plasmid-containing strain , but also o f the relatively rare R-plasmid conj ugat ive rec ip ient . Under this condition , the divers ity of drug-resistant bacteria ( i . e . , two d i fferent stra ins ) , as wel l as the extent o f res istance , i s increas ing . Although these conditions have been ne ither adequately modeled for potential analys is nor appropriately tested in an in vitro s ituat ion that would re flect actual forces in vivo , the theoretical cons iderations ra ise concern that subtherapeut ic concentrations of drugs may be doing as much harm as therapeut ic concentrations , i f not more , part icul arly in view o f the ir more cont inuous use . Veterinary studies discussed above lend credence to this concern . A complete cha in of direct evidence l inking human di sease caused by multiple-drug-resistant organisms to subtherapeutic use of penici l l in and the tetracycl ines in anima l feeds does not exist . Such evidence as does ex ist is l imited to outbreaks of multiple-drug-resistant salmonellosis . Conclus ive direct evidence of such a l inkage would include ful l characteri z ation of the infect ing sa lmone l lae based on ava ilable techniques of plasmid analys is ; epidemiologic evidence impl icat ing a part icular food ; isolation of the infect ing organism from the impl icated food and proof of its clona l ident ity ; epidemiologic evidence l inking the contaminated food with a particular farm service ; isolation of the infect ing organism from the impl icated animal s or poultry with proof of its clonal ident ity ; and documentat ion of the subtherapeut ic use of penic i l l ins or tetracycl ines in feeds consumed by the impl icated animal s or poultry . It might st i l l be argued that multiple-drugÂ res istant sa lmonel lae were present be fore subtherapeutic use of penic i l l ins or tetracycl ines was initiated , but most scient ists would accept the outl ined cha in of ep idemiologic
18 3 and microbiologic evidence as providing direct and conclus ive proof o f a cause-and-effect association . In only one outbreak , reported by Spika et al . , l 5 was this cha in o f transmiss ion ful ly documented ; the antib iot ic used in this instance was not penicil l in or tetracycl ine , but rather chl oramphenicol . The use o f chloramphenicol as a feed additive has never been approved by FDA in the United States , although in this instance it was used therapeut ical ly . Al l other reported outbreaks that impl icated multiple-drugÂ res istant salmonellae from an animal or farm source fal l short in their provision of evidence that conclus ively l inks the source of the drug-resistant organi sm with subtherapeutic use of antibiot ics in animal feeds . Those reports did not document ful ly the cha in of transmission , prove identity o f the infect ious salmonellae with those from the impl icated farm source , and document the type or amount of antibiot i c used in the animal feeds . The last has been part icularly dif ficult to ascertain in most of the disease outbreaks , i . e . , to estab l i sh retrospectively the precise antibiot ics or the amount used in feed . Thus , the studies of outbreaks of multiple-drugÂ res istant salmonellosis in humans , although they are the best evidence available , have not provided direct evidence of the human health risks due to subtherapeutic use of penic il l in or the tetracycl ines in animal feeds . LQSS OF DRUG BESISTAHCE Upon cessation of drug use , there should be a measurable and continuous decl ine in the concentration of drug in the environment . At a point at which thi s concentration is significantly below the MI C of the susceptible stra ins , those stra ins should mani fest a growth advantage over otherwise ident ical bacteria that in addition possess R plasmids . This advantage should be in direct relationship to the amount of diverted energy and raw materials the cel l needs to keep the R plasmid on board ( i . e . , new DNA , RNA , and protein synthesis ) and may be subtle . In sharp contrast to the drugÂ l oaded environment , where the effect of the drug on the populat ion of susceptible bacteria is seen within hours or days ( because of the enormous growth advantage of the res istant bacteria ) , the effect of antimicrob ial removal may take months or even years to be mani fested ful ly . The more subtl e the growth advantage of susceptibl e bacteria in the drug- free environment , the longer the period before the outgrowth of susceptible ( i . e . , R-plasmid- free ) bacteria can be seen . For these reasons any analys is of the e ffects o f drug remova l from the environment must b e extended past the immediate postwithdrawal period .
18 4 The prospect ive C DC studies of salmonellosis i n selected urban and rural counties showed that the overal l frequency of res istance to one or more antimicrobials had increased from 1 6 % to 2 4 % between 1 9 7 9 and 1 9 8 4 . 3 , 9 However , in one serotype , ï¿½ heidelberg , the frequency of res istance decl ined from 6 7 % to 3 5% during the 5-year period . Poultry was a common reservoir of ï¿½ heidelberg ; from 1 9 7 9 to 1 9 8 3 , CDC reported 6 9 % of the nonhuman isolates of ï¿½ heidelberg from this source . By the late 1 9 7 0 s , most poultry producers had stopped us ing penic i l l in and tetracylines as growthÂ enhancers ; for 1 9 7 9 - 19 8 2 , only 4 % of . broiler-chicken producers were reported as using low doses of the tetracyl ines in feed . The decl ine of res istance in this salmonel la serotype assoc iated temporal ly with the decrease in use of penic i l l in and the tetracyl ines as growth-enhancers suggests that decreased ant imicrobial res istance might fol low reduced use of these drugs in subtherapeutic dosages . However , whether the use of penic il l in and the tetracyl ines for disease prevention also decreased during 19 7 9 - 1 9 8 2 is unclear . The number of isolates of ï¿½ heidelberg studied was sma l l ; a much larger group of isolates should be examined to establ ish the val idity of this interesting prel iminary observation . In practice , the indicat ions for the subtherapeutic use of ant imicrobials for di sease prevention appears to the committee to be interpreted broadly . The goal o f such use might be to halt the spread of overt disease that has appeared in a few members of a herd . It appears to be used at certa in periods in the rearing of farm animals when they are cons idered to be particularly vulnerable to various infect ions ( e . g . , sh ipping- fever complex when cattle are moved into feedlots , and resp iratory diseases in pigs ) . Often , subtherapeut ic dosages are employed in feed for long periods without clear indicat ions . In the case of swine , they are used regularly at spec i fic stages of production : starter , grower , lactat ion , breeding , and gestation . Some farmers may be using ant imicrobial-containing feeds without be ing aware of it . Mixing procedures may be such that the concentrations achieved may exceed those targeted . Although distinct ions have been made between use of subtherapeutic doses of antimicrobials in feed for growth promotion and for disease prevention the value of dist inguishing between these two uses is rendered uncerta in by many aspects of current pract ice . It seems most rea sonable , there fore , to cont inue to categori z e both uses as subtherapeut ic , as they are currently viewed by the FDA . Better de fined guidel ines for use of subtherapeutic concentrations of ant imicrobials for disease prevention would be of bene f it .
185 FOREIGN EXPERIENCE IN BANHING ANtiBIQTICS AS FEED APDITIYES The Swann committee report of 1 9 6 9 in England addressed the issue of feed ant ibiotics ( subtherapeutic use ) and their effects on the selection of strains o f bacteria res istant to antimicrobial drugs . 1 3 It recommended that a l l ant imicrobials used in humans b e prohibited from use for growth promotion in animal s . It also stated that antimicrobial drugs used for humans could be used in treating animal s for disease or prophylactic indications when prescribed by a veterinarian . I n subsequent years , the central veterinary laboratory regularly conducted antibiot ic-susceptibil ity testing on stra ins of salmonel lae submitted to it . Its intention was to determine i f the Swann committee mandates influenced the antimicrobial susceptibi l ity patterns . It col lected data on trends of drug-resistance patterns over the years 19 7 2 , 19 7 4 , 197 6 , 1 9 7 7 , 1 9 8 4 , 198 5 , and 19 8 6 . Two maj or observations were based on these data . First , res istance patterns pers isted throughout the period : rarely was there any decrease . Second , one group of related phage types o f ï¿½ typhimurium ( 2 04 C was the predominant type ) appeared in calves in 1 9 7 9 . These stra ins are multiple-drug-resistant and are responsible for the increase in res istance patterns detected during this period . In 19 8 5 , 2 04 C constituted 6 2 % o f salmonella stra ins isolated from cattle . Almost a l l stra ins (more than 8 9 % ) were res istant t o tetracycl ines , ampici l l in ( and related penicil l ins ) , trimethoprim , and chloramphenicol . Resistance to gentamicin has gradual ly increased . ï¿½ dublin stra ins during these surveys were the second most common isolates a fter ï¿½ typhimurium . Most ï¿½ dublin were isolated from cattle . In 19 8 5 , less than 1% were res istant to tetracyc l i ne , ampici l l in , trimethoprim-sul fonamide , and chloramphenicol . Streptomycin and sul fonamide res i stance was more common-- 66 . 7 % and 2 8 . 6% , respectively . Most other salmonella serotypes isolated from cattle were susceptible to these antibiotics . In poultry , selected antimicrobial drugs demonstrated fa i l ure to inhibit growth of salmonel lae : 2 4 % of stra ins were res istant to streptomycin , 8 % to tetracycl ine , 7 2 . 8 % to sul fonamide at s o ï¿½g , and 1 1 % at 5 0 0 ï¿½g sul fonamide , and 0 . 8 % to chloramphenicol . It is clear that the phage type 2 04 C of ï¿½ typhimurium is an example of an unusual stra in that can periodica l ly cause epi z ootics . In 1 9 6 4 - 19 65 , an outbreak o f ï¿½ typhimurium phage type 2 9 occurred in cattle . By 1 9 6 9 , this epidemic was largely over . That stra in may have been selected through antibiotic pressure . The pers istence of it and phage type 2 0 4 C appears to be related to biologic
186 properties that permit intestinal colonization and abi l ity to induce disease . Strain 2 04 C has the propens ity to acquire plasmids . It probably became a problem in calves because o f multiple exposures associated with the many t imes when these anima l s were transported from broker to broker . The qreat mob i l ity of calves amonq brokers was di fferent from the s ituation in the years before the Swann committee recommendations . The use of antibiotics in animal husbandry in Enqland has not decreased , but rather has continued to increase . Thi s increase is due to prophylactic and therapeutic uses . Penici l l in and the tetracycl ines continue to be the most widely used druqs . This fact suqqests that they have not lost the ir e ffectiveness for treatinq animal diseases . Althouqh nonprescription uses of antimicrobials have been documented by the British Veterinary Association , the hiqher concentrations of antimicrobials in prescription-authori z ed therapeutic and prophylactic uses are blamed for selectinq res istant stra ins of salmonel lae and other bacteria in anima l s . The short l i fe span of the food anima l s and the apparent rapid decl ine in the number of res istant stra ins once the antimicrobial druqs are withdrawn are thouqht to be relatively ef fective barriers to a more widespread dissemination of these potential pathoqens . The incidence of salmone l l os i s in humans in Enqland has shown a persistent yearly increase from 19 7 0 to 19 8 0 . I n the subsequent four years , the incidence appeared to increase rapidly presumably because of increased numbers of ï¿½ typhimurium cases . Durinq these years other serotypes appeared , increased to a peak , and then usua l ly subsided to l ow numbers . Reasons for these variations are unknown . The isol ates of ï¿½ typhimurium phaqe type 2 04 C from humans have not had the same hiqh incidence of res istance to ant imicrobial druqs as those from animals . In 1 9 8 5 , 2 07 human isolates were tested and more than 9 2 % were susceptible ; in the same laboratory , of 1 , 0 5 0 bovine isol ates , only about 2 0 % were susceptible . The 2 07 human isolates represented 4 % of a l l ï¿½ typhimurium isolated in 1 9 8 5 , while 50% of the bovine stra ins were o f phaqe type 2 04C . Strain 2 04 C may be in the food chain , but it has not evolved in the same fashion as have the bovine stra ins ; it i s less common and has not devel oped the same hiqh inc idence o f res istance t o antimicrobial druqs . The data indicate that the Swann committee recommendations have not had a s iqn i f i cant e f fect on the number of resi stant stra ins of salmonel lae . This may be an unfair assessment , because there is no orqani z ed data base from be fore the recommendations with which to compare data col l ected later . Furthermore , some chanqes in aqricultural practices have occurred which have enhanced the spread o f salmonel lae .
187 Althouqh annual mortal ity rates in humans associated with salmonel losis in Enqland were not ava i lable to the committee , there were reports of deaths in various outbreaks , but deta i l s were not obta ined . However , there is no evidence of an increasinq mortal ity rate , as miqht be anticipated with an increasinq incidence of infections . THE RISK MODEL The committee learned that a s imilar risk model had been used by the National Resources Defense Counci l ( NRDC ) in its pet it ion dated 2 0 November 19 8 4 to the Secretary of Health and Human Services askinq for suspension o f the approval o f the new animal druq applications for subtherapeutic use o f penici l l in and the tetracycl ines i n animal feeds . l 6 The NRDC a l l eqed that the use of these druqs presented an imminent hazard to the public health . The committee ' s risk model and the parameter estimates used in it are summa rized in Table VI I I - 1 . The NRDC used "best estimates , " whi l e the committee used three estimates : l ow , mid-ranqe , and h iqh and appl ied these to five distinct parameters in the risk model . In comparison , the NRDC estimate o f the number o f deaths per year due to salmonel los is associated with subtherapeutic use of penici l l in and the tetracycl ines was 1 1 6 deaths , that corresponds to this committee ' s mid-ranqe estimates in Table IX- 1 of 3 0 deaths--a rather s imilar result in the face of so much uncerta inty . The second NRDC estimate , 2 6 4 deaths per year , is based on a d i fferent method that starts from an estimated 1 , 0 0 0 deaths per year due to salmonel losis , a fiqure we bel ieve to be too hiqh . The Food and Druq Administration ( FDA) , a constituent aqency of the Department of Health and Human Services ( DHHS ) , careful ly analyzed the NRDC petition and recommended that the Secretary of DHHS deny the petition on the qrounds that an " imminent hazard " had not been demonstrated . 17 The pet ition was in fact denied . The FDA ' s analysis concluded that the NRDC had not shown in its petition that ant ibiotic res istance caused by the subtherapeutic use of penic i l l in or the tetracycl ines in animal feed had a s iqni ficant impact on the outcome of a siqnificant number of cases of salmonel losis and thus , that no " imminent hazard" had been demonstrated . The FDA ' s analysis first discussed the d i f f iculty of treat inq infections by resistant salmone l l ae . I t did not accept arquments about increased d i f ficulty in treatment , because most infections with sa lmonel l ae are uncompl icated and resolve without treatment ( so " antibiotic treatment is not recommended in patients with the uncomplicated diarrheal type of salmonel losis , " and for those cases occurrinq outs ide the intestine the druq of choice is chl oramphenicol , to which
188 only about 0 . 7 % of salmonel lae are res istant and for which alternative drug therapies exist ) . The FDA a l so concluded that the data then ava ilable did not demonstrate any alteration in virulence and contended that some salmonella deaths ( such as from heart attacks caused by dehydration and stress due to salmonellos i s ) are unrelated to antibiotic therapy . The FDA then cons idered what we cal l the " et iologic fraction , " as wel l as inappropriate therapy for infections not at first recognized as salmonel losis , and concluded that neither had been shown to present a maj or problem . Fina l ly , the FDA commented on the NRDC parameters ( see Table IX- 1 ) and took special issue with the estimated death rate of 4 . 2 % , on the grounds that it was subj ect to a number of potential b iases and l imitations , including lack o f documentation that salmonel losis was the primary cause o f the repo rted deaths . ( The largest dif ference between our midÂ range estimate and the NRDC estimates is in the death rate . We queried the CDC , as noted above , and found that some deaths were indeed not due to salmonellosis and that others were questionable : we reduced our mid-range estimate accordingly) . The FDA also concluded that the NRDC ' s estimate of 69% of res istant salmonel lae traceable to animal sources was based on a very l imited sample and that such deaths caused by subtherapeutic use of penici l l in or the tetracycl ines ( estimated by NRDC as 5 0 % ) could not be est imated accurately from the ava ilable data . The committee has considered these obj ect ions careful ly , in l ight of advances in scient i fic understanding s ince 1 9 8 4 and the whole body o f data ava ilable at the time that it worked on this matter ( the first hal f of 1 9 8 8 ) . The committee has not tried to j udge the merits of either the NRDC ' s petition or the FDA ' s response . The committee bel ieves , however , that some estimates can be made , as shown in Figures VII I - 1 through VI I I - 1 2 . These estimates are sti l l highly uncerta in , as indicated in the figures themselves . EVIDENCE SUGGESTING THE PRESENCE OF HAZARD The estimates presented here have wide margins o f poss ible error , a s reflected i n the ranges from the 5th to 9 5th percenti les ( percent i l e is the scriptor for fraction of estimates fal l ing bel ow it and are not confidence l imits ) in Figures VI I I - 1 through VI I I - 1 2 . This is a direct reflect ion of the compounding of estimates of component factors that themselves have substantial ranges from the lowest plausible to the h ighest plausible estimate . I f our model is to be adopted for future use , we urge that the respons ible authorities promote the appropriate research to produce the data needed to narrow each of the ranges of estimates shown
189 TABLE IX- 1 COMPARISON OF PARAMETERS IN THE PRESENT REPORT WITH THOSE OF THE NRDC REPORT : HUMAN SALMONELLOSI S DEATHS ATTRI BUTABLE TO ANY LOW-LEVEL FARM USE OF PENICILLIN OR THE TETRACYCLINES Present Report NRDC Report (Mid-Range ( Fi rst Estimates > Model l Reported cases per year 50 , 000 4 0 , 000 in the u . s . Res i stance to penicill in 0 . 15 0 . 20 and/or tetracycl ines Death rate for salmonellosis 0 . 008 0 . 04 2 from res istant stra ins Fraction associated with 0 . 70 0 . 69 * stra ins o f farm origin Fraction caused by 0 . 50 0 . 72 subtherapeut ic use o f penici l l in and/or the tetracycl ines Product of these estimates : 30** 116 Deaths per year in the u . s . from sa lmone l l ae res i stant to penici l l in and/or the tetracycl ines because o f subtherapeut ic use of these drugs on the farm . Source : Prepared by the committee using data from Table VI I I - 1 and from the Natural Resources Defense Counc i l . 1 6 * " Traceable t o animal sources . " ** Di f fers s l ightly from Table VI I I - 2 , because the product of the mid-range est imates is not necessarily the med ian o f the 2 4 3 products .
190 i n Table VI I I - 1 , and thereby substantial ly reduce compos ite ranges of estimates shown in Figures VI I I - 1 through VI I I - 12 . The presently available data are an incomplete " patchwork" from a variety of sources ; they are not col lected systematica l ly for the nation , they are complex , they are frequently of poor qual ity and require extrapolation for use in risk assessment , and they are not focused on the spec i fic points o f direct interest . These characteristics o f the ava i l able data are inherent in the problem of col lecting data and are not the fault of any one government agency or researchers who have studied this problem over the past several years . For example , none of the sources summa ri z ed in Table VI I - 8 was focused on estimation of the " etiologic fraction , " none presented an est imate of that fract ion , each had very sma l l samples for this use , and each was subj ect to substantial bias in the identi fication and recruitment o f subj ects . s imilarly , there have been few opportunities for the accurate and unbiased estimation of population-wide death rates , though countless reports of salmone l l ae deaths and death rates have been publ ished for use in other contexts . We regard the model itsel f as neutral - -this is , unbiased with respect to errors or uncertainties in the estimates it produces--though of course it re flects a l l the errors or uncerta inties that are inherent in the parameters it uses ( Tabl e VI I I - 1 ) . Although the model itsel f is neutral , it can perhaps be improved , especially with respect to the path impl ied by the column headings of Table IX- 1 , the number of steps (which we took to be f ive ) , and mod i f i cations to make better use of ava i l able data . We invite and urge others to prepare alternative mode l s , and we hope that funding agencies and sponsors of research in this field wi l l increase the ir support o f e fforts to devel op improved mode l s . REFERENCES 1. Anderson J . D . , w . A . Gil lespie , and M . H . Richmond . Chemotherapy and ant ibiotic res istance trans fer between enterobacteria in the human gastrointestinal tract . J . Med . Microbial . 6 : 4 6 1 - 4 7 3 , 1 9 7 3 . 2. Achtman , M . , M . Keuzenroeder , B . Kusecek , H . Ochman , D . caugant , R . K . Sel ander R . K . , v . Va isanen-Rhen , T . K . Korhonen , s . Stuart , F . Orskov , and I . Orskov . Clonal ana lys is of Escherichia col i 0 2 : K1 isolated from di seased humans and anima l s . Infect . Immun . 5 1 : 2 6 8 - 2 7 6 , 198 6 .
19 1 3. Cohen , M . L . , and R . v . Tauxe . Drug-resistant Salmonella in the United States : An epidemiologic perspective . Sc ience 2 3 4 : 9 6 4 - 9 6 9 , 19 8 6 . 4. Gardner , P . , D . H . Smith , H . Beer , and R . c . Moe l lering , Jr . Recovery of res istance (R) factor from a drug-free community . Lancet 2 : 7 7 4 -7 7 6 , 1 9 6 9 . 5. Humme l , R . , H . Tschape , and w . Witte . Spread o f plasmid-mediated nourseothricin resi stance due to ant ibiotic use in animal husbandry . J . Basic Microbiol . 2 6 : 4 6 1-4 6 6 , 19 8 6 . 6. Levy , s . B . , G . B . Fitzgerald , and A . B . Macone . Changes in intestina l fl ora of farm personnel a fter introduct ion of a tetracycl ine-suppl emented feed on a farm . N . Engl . J . Med . 2 9 5 : 5 8 3 - 5 8 8 , 19 7 6 . 7. Linton , A . H . , K . Howe , P . M . Bennett , M . H . Richmond , and E . J . Wh ites ide . The colon i z ation o f the human gut by ant ibiotic res istant E scherichia Â£Qli from chickens . J . Appl . Bact . 4 3 : 4 6 5-4 6 9 , 1 9 7 7 . 8. Linton , A . H . Antibiotic res i stance : The present s ituation reviewed . Vet . Rec . 1 0 0 : 3 54 - 3 6 0 , 1 9 7 7 . 9. MacDonald , K . L . , M . L . Cohen , N . T . Hargrett-Bean , J . G . Wel l s , N . D . Puhr , s . F . Col l ing , and P . A . Bl ake . Changes in antimicrobial res istance of Salmonell a isolated from humans in the United States . J . Amer . Med . Assoc . 2 58 : 1 4 9 6 - 1 4 9 9 , 1 9 8 7 . 10 . Pace , w . E . Food Contamination . In National Research Counci l , The E f fects on Human Health of Subtherapeut ic Use of Antimicrobia l s in Animal Feeds . Appendix H . Washington , D . C . : National Academy Press , 19 8 0 . 11 . Parsonnet , K . c . , and E . H . Kass . Does prol onged exposure to antibiotic-resistant bacteria increase the rate of antibiotic-res istant infection? Antimicrob . Agents Chemother . 3 1 : 9 1 1-9 14 , 1 9 8 7 . 12 . Petrochei l ou , v . , J . Grinsted , and M . H . Richmond . R plasmid trans fer in vivo in the absence of ant ibiotic selection pressure . Antimicrob . Agents Chemother . 1 0 : 7 5 3 -7 6 1 , 19 7 6 . 13 . Swann , M . M . , et al . Joint Committee on the Use of Antibiotics In Animal Husbandry and Veterinary Med icine . London , England : Her Maj esty ' s Stationery O f fice , 1 9 6 9 .
19 2 14 . Smith , M . G . In yiyo transfer o f an R factor with in the lower gastrointestinal tract of sheep . J . Hyg . 7 9 : 2 59 - 2 68 , 19 7 7 . 15 . Spika , J . s . , s . H . Waterman , G . w . Soo Hoo , et . al . Chloramphenicol-res istant Salmonella newport traced through hamburger to da iry farms . N . Engl . J . Med . 3 1 6 : 565-57 0 , 1 9 8 7 . 16 . u . s . Food and Drug Administration . The National Resources Defense Council , Inc . submiss ion of a petition to the Secretary o f Health and Human Services . Fed . Reg . 4 9 ( 2 4 7 ) : 4 9 64 5- 4 9 64 7 , Friday , December 2 1 , 1 9 8 4 . 17 . u . s . Food and Drug Administration . Recommendation on Imminent Hazard Petition Subtherapeutic Use o f Penicil l in and the Tetracycl ines i n Animal Feeds ( filed with hearing clerk in response to 19 8 4 NRDC submiss ion o f petition to DHHS ) .
X CONCLUS IONS The committee has reviewed the extens ive and sometimes confl icting l iterature pertaining to poss ible human health r isks associated with the use o f subtherapeutic concentrations o f penici l l in and the tetracycl ines ( and other antimicrobial s ) in animal feed . It evaluated invest igations of the mol ecular nature of plasmids , transposons , and other bacterial antimicrobial-resistance determinants and thei r trans fer 1 data o n the extent of antimicrobial res i stance i n Salmonel la species ( and in other enteric pathogens ) isolated from humans and farm anima l s 1 epidemiologic studies in humans and farm anima l s ; data on reported cases of human i l lness and deaths due to Salmonel la transmitted to humans from farm animal s via meat and poultry products ; in format ion on the extent o f subtherapeutic use of penic i l l in , the tetracycl ines , and other antimicrobials in animal feed ; and data from Great Brita in on the ef fects of the restrictions placed some years ago on the use of antimicrobials in animal feeds in those countries . The committee also reviewed the ava i l able publ i shed reports deal ing with four subj ects recommended for further study in the 1 9 8 0 report of the National Research Counci l Committee t o Study the Human Health E f fects of Subtherapeutic Antibiotic Use in Anima l Feeds : the e ffects of subtherapeutic and therapeutic doses o f antimicrobials on the preval ence of antimicrobial -resistent enteric bacter ia ( including salmone l l a e ) in farm animals ; the extent o f carriage o f res istance- factor-containing bacteria i n vegetarians and nonvegetarians ( to ascerta in the extent to which such carriage is associated with meat consumption ) ; the extent o f carriage of res istance- factor-conta ining Enterobacteriaceae in abatto ir workers , their fami l ies , and neighborhood control s ( to assess the association with occupationa l exposure to bacteria from an ima l sources ) ; and the preva lence of urinary tract infections ( and urinary tract in fect ions due to res i stance-pl asmid-conta ining Enterobacteriaceae ) in fema le workers in poultry-process ing plants and a control group of women without contact with farm animals or the ir unprocessed meat products . We consul ted with and heard testimony from the epidemiol ogy sta f f of the Centers for Disease control , other med ical ep idemiologists , veterinarians , representat ives of 193