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.
194 the Animal Health Institute , microbiologists , and representatives of the pharmaceutical industry . Us ing a l l the resources noted above , we were unabl e to find a substantial body of direct evidence that establ i shed the existence of a def inite human health hazard in the use o f subtherapeutic concentrations of penici l l in and the tetracycl ines in animal feeds . However , we bel ieve that important--but as yet scant--data indicate the flow o f distinct salmonel l a strains from farm anima l s , through the food process ing cha in , to humans in whom they cause cl inica l salmone l l osis . In the one compel l ing instance o f such a clear l ink , the multiple-antibiotic-resistant ï¿½ newport originated in farm animals exposed to chl oramphenicol , a drug not approved by the Food and Drug Administrat ion for use in feed . The committee bel ieves that the molecular f ingerprinting techniques used in this study can provide (when unique markers are present ) the direct evidence needed to trace the source of antibacteria l - resi stant bacteria to human in fection . I f records of amounts of antibiotic use are ma inta ined on farms producing food for human consumption , better evidence can be estab l i shed for incriminating subtherapeut icjtherapeut ic doses in disease outbreaks . The committee bel ieves that there is indirect evidence impl icating subtherapeutic use of antimicrobia l s in producing res i stance in infectious bacteria that causes a potential human health hazard . The ev idence is of several kinds : o There are extens ive experimental data on the propert ies of R pl asmids and the ir capacity for trans fer o f antimicrobial-resistance determinants , both i n the test tube and in the intestinal tract , particularly in the presence of antimicrobial sel ective pressure . o There is evidence of widespread use o f subtherapeut ic concentrations of pen ic i l l in and the tetracycl ines ( and other antimicrobia l s ) on farms and feedlots . o There i s ample evidence o f high leve l s o f antimicrobial res i stance among animal isol ates of sa lmone l l ae . o Animal and poultry carcasses in meat-process ing pl ants are o ften contaminated with Escher ichia coli and other enteric pathogens . Few data are ava ilable on the frequency of antimicrob ial res istance among such isol ates . I f the preva lence of antimicrob ial res istance among reported isol ates from d iagnostic laboratories is a true representation of antimicrobial resi stance in farm anima l s going to sl aughter , the frequency of resi stance among enteric pathogens in animal and poultry carcasses would be expected to be high . However , if the salmone l l a isolates reported
195 f rom diagnostic laboratories are principal ly from anima l s that are i l l and have received antimicrobial s , the figures would clearly overestimate the frequency of resistant i solates from meat and poultry carcasses . o Handl ing and ingestion o f improperly cooked , packaged f ro z en or refrigerated meat and poultry contaminated with bacterial pathogens provides exposure to an in fecting i noculum . o Experience with antimicrobial drugs in humans over the last 4 5 years has revealed the emergence o f res i stant stra ins associated with extens ive drug use and the need to avoid unnecessary and prolonged use , particularly "prophylactic" use without clear and proven indications . I n addition , the committee has used the results provided by the risk assessment model presented to estimate quantitatively the poss ible risk of mortal ity associated with ant ib iotic-res istant salmone l lae due to the subtherapeutic use o f penic i l l in or the tetracycl ines in anima l s . In the 1 9 8 0 NRC report , the committee to study the Human Health E ffects o f Subtherapeutic Antibiotic use in Animal Feeds concluded that " the postul ated ha zards to human health from a subtherapeut ic use o f antimicrobials in animal feeds were ne ither proven nor di sproven . " In other words , the risk o f human health as a result of subtherapeutic use o f antimicrobials i n feed was not estimated . We found the ava i lable data base on some aspects of the problem to be l imited in qual ity and quantity : indeed , the data had not been gathered prospectively for the purpose of thi s type o f analys is . The committee has used what it cons iders the best ava i lable information , indicating , where appropriate , the inherent weaknesses in the data . Admittedly , in some instances , we used only the best estimates ava i l able in the risk assessment . The assessment does indicate the presence of risk . Although it does not provide a distinct numerical " answer" to the quest ion of the magnitude of the human hea lth risk involved , it does provide some indication o f the probable s i z e of the risk in terms o f numerical estimates o r ranges . These are presented bel ow a s numbers o f deaths per year attributabl e t o the subtherapeutic use o f antimicrobials ( or penici l l in and tetracycl ine s ) in the l i sting of speci fic conclus i ons : BIOLQGIC IMPACTS o Use of each new antimicrobial agent over the last hal f-century has eventual ly mobi l i zed genes that encode res istance to the agent and disseminated them widely through
196 the world ' s interconnectinq bacterial populations . U s e o f the ant imicrobial aqent disseminates the res istance qenes in staqes , each of which beqins with a rare molecular event that faci l itates further dissemination . Althouqh use o f ant imicrobials i n a patient or the patient ' s neiqhbors miqht have triqqered overqrowth and cl inical mani festat ion of the res istant strain , the evolution and del ivery o f its res istance qenome was the result of prior use in many , probably distant , bacterial populations . o Results of surveys of isolates of salmonel lae from animals and humans in the United States and restrictionÂ endonuclease fraqment patterns of res istance plasmids from selected isolates suqqest that clones of resistant salmonel lae are endemic in animal s and sporadic or occas ional ly ep idemic in humans . o Herds of farm an imals qiven subtherapeutic amounts of antimicrobial aqents have more antimicrobial -res istant intestinal bacteria than herds qiven no antimicrobial s . o The most important determinant in the selection o f ant imicrobial -resistant stra ins i n a bacterial population i s exposure of that population to antimicrobial s . Tota l duration and concentration of antimicrob ial use are important in selection for res istance . Any measure that fa i l s to reduce tota l use apprec iably is unl ikely to a ffect the preval ence of antimicrobial-resistant stra ins . o Res istance to ant imicrobial druqs amonq sa lmonel l a stra ins can interfere with the eff icacy of antimicrobial therapy o f human salmonel losis . ( Such res istance i s usual ly R-plasmid-mediated , so it can involve other druqs , such as trimethoprim-sul famethoxa zole , chl oramphen icol , and ampicil l in . ) Althouqh such interference with the e f f icacy of therapy almost certa inly occurs ( i . e . , patients are treated with an antimicrobial that is ine f fective because of druq res istance ) , it is probably quite uncommon in nontyphoidal sa lmonel los i s . o The ava i l able data are inadequate to conclude that either subtherapeutic or therapeut ic concentrat ions o f antimicrobials are more sel ect ive of druq-resistant bacteria . On theoretical qrounds , it is l ikely that therapeut ic and subtherapeut ic dosaqes exert equal sel ective pressure for cl onal expans ion of res istance , but subtherapeutic dosaqes exert more pressure for conj uqative trans fer of druq res i stance , because of the dosaqes and the durat ions o f administrat ion .
197 o An imal and poultry products ( includinq veal , bee f , pork , chicken , eqqs , and milk) are the princ ipal sources o f human nontyphoidal salmonel losis . Also , some ï¿½ QQli serotypes can also be found in the intestinal flora both o f humans and of farm animals . Thus , there could be an interconnect inq l ink between these two larqe pools o f enteric m icroorqanisms , facil itated by the h iqh frequency of contamination o f anima l and poultry carcasses in s lauqhterhouses . Such a potential l ink would provide a means of movement of R plasmids of farm oriqin to the human a l imentary tract . The interconnection , because of its nature , would constitute an almost exclus ively one-way passaqe . o The overal l prevalence of res istance to any of five commonly used antimicrobials is about 4 times as qreat in col lections of salmonella isolates from farm anima l and poultry ( 6 5 % ) as those in col lections of isolates from humans ( 15 . 5% ) . This d i fference suqqests that the predominant pool of res istant salmonel lae is in farm animals . Because ult imately almost a l l human infections with nontyphoidal salmonel lae result from stra ins oriqinat inq in farm animal s , the ant imicrobial res istance observed in human isolates most l ikely is derived from the anima l pool of res istance qenes , rather than from selection due to antimicrobial use in humans . EPI DEMIOLOGIC FINDINGS o Evidence is sparse that directly l inks the use of penici l l in and tetracycl ine in subtherapeutic concentrations in animal feeds to human infections . several studies have yielded rel iable evidence of spread , from farm animal s and poultry to humans , of ï¿½ col i stra ins in which antimicrobial res istance had been induced by administrat ion of subtherapeut ic concentrations of antimicrobials as feed add itives . There is evidence from only one study of the direct spread of multiple-antimicrobial-res istant salmone l l ae from farm animals to humans via meat products . However , the ant imicrobial used on the farm was chl oramphenicol , a druq not approved by FDA as a feed addit ive in animals used for food production . It miqht be diff icult , or imposs ible , to provide a total cha in of evidence directly relatinq the maj ority of cases of human infection with antimicrobialÂ res i stant salmonel l ae to a source on the farm or feedlot or to relate the presence of the res istance to the use of spec i f ic ant imicrobials in subtherapeutic concentrations in feed . By the t ime a deta iled invest iqation o f an outbreak of human salmonel losis occurs , evidence of prior antimicrobial use patterns miqht not be ava i lable .
198 o It has not been poss ible to determine whether antimicrobial res istance of salmone l l ae caused by the administrat ion of subtherapeutic concentrations o f antimicrobials i n animal feed increases the number o f cases of human salmonel losis . o Whether the presence of antimicrobial res istance in salmone l l ae increases virul ence i s uncerta in ; the avai l able data are l imited and confl icting . I n special circumstances , as when R plasmids are l inked with virul ence genes ( e . g . , those for enterotoxin or hemolys in in ï¿½ ï¿½) , selection by antimicrobial agents might promote spread of v i rulent stra ins ; however , such an occurrence has only rarely been reported . It is not clear whether the overal l prevalence o f salmonel l ae i n food products i s increased b y v irtue o f antimicrobial res istance . However , the incidence of human sa lmone l l o s i s in the United States is increas ing , and the increase is unl ikely to be an arti fact of better reporting . As long as most stra ins of Salmonel l a are susceptible to the antimicrobials to which they are exposed , subtherapeutic administration of antimicrobials might reduce the preval ence of salmonel l ae in meat and poultry products that humans ingest . However , as the preval ence of res i stant stra ins increases because of repeated and prol onged exposure to antimicrobial s , subtherapeutic administration might actual ly favor the increase by suppress ing the normal compet ing flora and promoting R-plasmid spread . Direct proof of this pattern in sa lmone l l ae in farm animals is lacking . o The current frequency of R-plasmid-mediated ant imicrobial res istance among isolates of ï¿½ QQli and salmone l l ae in the intestinal contents of farm animals and poultry is h igh--much higher than in human isolates . It would be d i f f icult to predict the period required , after curta i lment of the use of subtherapeutic concentrations o f penic i l l in and the tetracycl ines i n animal feed , for RÂ plasmid-mediated ant imicrobial res istance to decrease in any extent in salmonel lae and ï¿½ ï¿½ strains . Maj or decreases might occur only a fter the passage of years , in view o f ( 1 ) the current degree of res istance , ( 2 ) the extensive environmental contaminat ion on farms and feedlots with res istant organisms , ( 3 ) the prolonged prior subtherapeutic use of antimicrobial s , which has a l l owed extens ive permeation of res istance genes ( transposons ) throughout the h ighly col on i z ation-adapted col i form fl ora of farm animals , and ( 4 ) the need to introduce competing , ant imicrobialÂ suscept ible col i form bacteria . Results of studies in conf ined populations of swine indicate that it could take many years for maj or decreases in level s of res istance to occur .
199 o Althouqh the extent of antimicrobial res istance amonq salmone l l a strains isolated from humans is probably qrowinq , it i s sti l l low enouqh for suitable intervention to forestal l pos s ible further increases and eventual ly t o lower the overa l l extent of antimicrobial res istance . ANtiBIOTIC USE PATTEBNS o The use o f subtherapeutic dosaqes of penici l l in and the tetracycl ines in animal feeds is extens ive in the United states . Such use is for the purpose of e ither qrowth promotion or disease prevention and often continues for a substant ial port ion of the qrowth cycle of farm animal s . The spec i fic rationale for use in a qiven herd at a qiven time is not always clear . Of over 3 1 m i l l ion pounds o f antimicrobials produced each year i n the Un ited States , about 4 2 -4 8 % is des iqnated for addition to anima l feeds or other unspeci f ied (minor ) uses . The best estimates ( they are only est imates ) indicate that penici l l in and the tetracycl ines account for almost 6 0 % of the antimicrobial s sold to the feed trade ( and presumably ultimately used on farms and in feedlots ) . Of the total amount of tetracycl ines produced in thi s country , for use in both humans and animals , approximately 7 0 % i s sold for use in l ivestock and poultry feeds . An estimated 8 8 % of all antimicrob ial use in l ivestock and poultry i s in subtherapeutic concentrations . Thus , subtherapeutic use of pen ic il l in , the tetracycl ines , and other ant imicrobials in animal feeds --wh ich accounts for some 4 0% of antimicrob ial production in the United States-Â constitutes a s i z able seqment of the total antimicrobial selective pressure ( for res istant enteric microorqanisms ) exerted on the combined human and farm-animal intestinal bacterial populations . o Interpretation of the results in Great Brita in a fter banninq the subtherapeutic use of penici l l in and the tetracycl ines in animal feed is d i f ficult , in part because total farm use o f these antimicrobials miqht not have decreased because use could have taken the form of therapeutic or prophylactic doses in feed for disease treatment or prevention as prescribed by a veterinarian . The appearance of new epidemic stra ins of ant imicrobial -resistant salmonella serotypes durinq the period of interd iction of subtherapeutic use further confounds interpretation . It miqht take years for d ilution of antimicrobial -res i stant stra ins of Salmonella and ï¿½ ï¿½ in the farm animal population be fore any substantial chanqes miqht be observable .
2 00 RISK ANALYSIS o The committee has been unable to f ind substantial direct evidence that bacterial res istance resulting from the use o f subtherapeutic concentrations of penici l l in or the tetracycl ines in animal feed causes an excess risk to human health as a result of consumption of food products derived from the treated an imals , as a result of contact with such animals , or as a result of exposure to an env ironment contaminated by res istant enteric bacteria from such an imals . Lacking this direct evidence , the committee turned to the tools of risk assessment to develop some quantitative estimate of the probable risk to human heal th associated with this form of the subtherapeutic use o f these antimicrobial s . o Use of penicil l in and the tetracycl ines in subtherapeutic concentrations in animal feed has led to increased antimicrobial res istance in foodborne commensal s and pathogens . The risk analys is i n thi s report focused only on human infection with salmonel la serotypes , because ava i lable data on other species were insuffic ient . The committee has not assessed the potential risk to human health associated with drug res istance in other gram-negat ive bac i l lary species (Campylobacter jei uni , Yersinia enterocolitica , and enterohemorrhagic ï¿½ Â£Qli ) of animal origin , because the data on human cases are too l imited and because antimicrobial suscept ib il ity data on those bacteria are not routinely obta ined . o Because the committee ' s risk assessments are based on estimates us ing sparse data , these estimates should be interpreted and used with caution . Such estimates are best seen as scienti fic hypotheses about the possible extent of a problem . Th is does not mean that they are "hypothet ica l " in the weak sense of be ing speculative . Rather , they are hypotheses that are cons istent with a l l ava i l able informat ion and scient i f ic understanding , but they have not been tested by traditional scienti fic methods . Al l the est imates presented in this report should be viewed in that perspective . o Annual numbers of deaths from salmone l l o s i s attributable t o subtherapeutic uses of any antimicrobials for prophylaxis and growth promotion have been estimated . Â· The l ikel iest estimate is 7 0 deaths per year . o The l ikel iest estimate of mortal ity from salmonel los is attributable to subtherapeutic uses of penici l l in/ampicil l in andj or tetracycl ine for prophylaxis and growth promotion is 4 0 deaths per year . Caveat--these are not necessari ly " excess deaths , " but rather estimates of the
2 01 yearly mortal ity attributable to salmonellosis o f the indicated origin . The deaths might to some extent replace deaths ( in the same patients or others ) that occur from infections due to salmonellae susceptible to penici l l in/ ampici l l in and tetracycl ine i f subtherapeutic dosages of these antimicrobials had not been used in animal feed . Estimation of such " replacement" o f deaths is not poss ible with the evidence at hand . o The l ikel iest estimate o f mortal ity from salmonellosis attributable to subtherapeutic uses of any antimicrob ial for growth promotion only is 2 0 deaths per year . As in the preceding ( and fol l owing ) estimates , the caveat regarding " excess deaths " appl ies . o The l ikel iest estimate o f mortal ity from salmone l l osis attributable to subtherapeutic uses o f penicil l in/ampici l l in andj or tetracycl ine only for growth promotion is 1 5 deaths per year . o The l ikel iest estimate of mortal ity from salmonellosis in the " etiologic fraction " attributable to subtherapeutic uses of any antimicrobial for prophylaxis and growth promotion is 6 deaths per year . The " etiologic fraction " is the proportion of persons exposed to an antimicrobial-res i stant salmonel la stra in who are at increased risk o f i l lness by virtue of recent use of antimicrobial drugs for whatever reason . Therefore , such deaths can be cons idered as " excess deaths " ; i . e . , they would not occur if the in fecting salmonel l a strain were not ant imicrobial -res istant and i f its multipl ication were not promoted , presumably , by suppress ion of growth of the competing normal antimicrobial-susceptible normal f lora . In the same way , the number of foodborne pathogens ( inoculum s i z e ) needed to precipitate disease might have been decreased . Whether a s imilar e ffect can be produced by prior antimicrobial use in persons infected with antimicrobial Â susceptible salmonel lae ( due to poss ible d i fferential ant imicrobial susceptibi l ity between susceptible salmonellae and normal components of the intestinal flora ) is unknown , and the committee has not been able to f ind data bearing on this question . o The l ikel iest estimate of mortal ity from salmonel losis in the " etiologic fraction " attributable to subtherapeutic uses of penici l l in/ ampic i l l in andjor tetracycl ine for prophylaxis and growth promotion is 6 deaths per year . o The l ikel iest estimate of mortal ity from salmonel l os i s in the " etiologic faction" attributable to
2 02 subtherapeutic uses of any antimicrobial only for growth promotion is 2 deaths per year . o The l ikel iest estimate of mortal ity from salmonellosis in the " etiologic fraction" attributable to subtherapeutic uses of penic i l l in/ampic i l l in and/or tetracycl ine only for growth promotion is 2 deaths per year . o Infections with antimicrobial -res istant strains o f Salmonella are more o ften fatal than infections with susceptible Salmonella . Therefore , the increased d i f f iculty o f providing e ffective therapy for human disease can be estimated . The increased d i f ficulty in providing e ffective treatment may be due to increased virulence of ant imicrobialÂ res istant stra ins , to the presence of res istance to one o f the ant imicrobials ordinari ly used t o treat such infect ions when they are severe or when they occur in particularly vulnerable persons , or to some other factor . The l ikel iest estimate of mortal ity from salmone l l osis arising because of increased d i f ficulty of treatment attributable to subtherapeutic uses of any antimicrobial for prophylaxis and growth promotion is 4 0 deaths per year . o The l ikel iest estimate of mortal ity from salmonel losis aris ing because of increased d i f ficulty o f treatment attributable t o subtherapeut ic uses o f penic i l l in/ampici l l in and/ or tetracycl ine for prophylaxis and growth promotion is 2 0 deaths per year . o The l ikel iest estimate of mortal ity from salmonel losis aris ing because of increased d i f f iculty o f treatment attributable t o subtherapeutic uses o f any antimicrobial only for growth promotion is 8 deaths per year . o The l ikel iest estimate of mortal ity form salmonel l osis aris ing because of increased difficulty of treatment attributable to subtherapeutic uses of penici l l in/ ampici l l in and/or the tetracycl ines only for growth promotion is 8 deaths per year . o Evaluation of the foregoing estimates o f mortal ity from salmonel losis attributable to subtherapeutic uses of antimicrobials in animal feed requires cons iderat ion in a broader context . What possible bene fits accrue from such subtherapeutic use of antimicrobials in food product ion? Would human deaths from salmone l l os i s be reduced by the discontinuation of subtherapeutic use of penici l l in/ ampici l l in andjor the tetracycl ines? The committee ' s thesis is that , although some deaths due to antimicrobial -res i stant stra ins might be " replaced " by deaths due to suscept ible stra ins , the tota l number of deaths would decrease , however ,
203 this cannot now be proved . The committee o f fers no recommendat ions regarding pol icy-making because that was not part of its mandate .
XI RECOMMENDATIONS FOR FUTURE RESEARCH The committee offers no recommendations o f poss ible solut ions to risk manaqement of the overal l problem under cons ideration . It has directed its attention mostly to its charqe to review the human health consequences and the ri sk assoc iated with the use of penic i l l in and the tetracycl ines at subtherapeutic concentrations in animal feed . Recommendation of any action would be appropriate only a fter requlatory aqency review and we iqhinq of both the benef its and risks o f use of these antib iotics in subtherapeut ic dosaqes . The committee does , however , offer recommendations concerninq further investiqations that would be helpful in resolvinq the issue , which has been intensely debated for some 15-2 0 years . Many of the recommendations for study would remain appropriate whether current pol icies reqardinq subtherapeutic antimicrobial use rema in in e f fect or are chanqed by a requlatory aqency . In the former instance , the data obta ined would serve to strenqthen the informat ional underpinninq of risk estimation . In the latter instance , they would make it poss ible to compare data on the prevalence of antimicrobial res istance amonq enteric pathoqens and human health risks before and after institution o f any chanqe in approved antimicrobial use . STQDIES TO SUPPLEMJNT THE QATA 8ASE FOR RISK ANALYSIS The risk assessment performed by thi s committee used the best avai l able data rel ated to the six essential elements in its risk estimates : res istance of human salmonel l a isolates to antimicrobia ls , annual reports of cases of salmonellosis , death rates associated with antimicrobial-susceptibl e and antimicrobial -res istant stra ins , fraction of human salmonel losis deaths associated with stra ins o f farm oriqin , fraction o f antibiotic res istance in stra ins o f farm oriqin caused by subtherapeutic use of antimicrobials in anima l feed , and " etioloqic fract ion . " In compil inq its risk estimates the committee was l imited by the pauc ity of some types of data and the consequent need to extrapolate from the results of sma l l studies to the qloba l problem , by the fact that in some subj ects rel iable data were almost total ly lackinq and had to be substituted for with " best estimates , " 2 04