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.
2 the presence of a human health hazard that resulted from the use of subtherapeutic concentrations of penic i l l in and the tetracycl ines in animal feeds . Nonetheless , the committee bel ieves that important , although as yet sparse , data show the flow of distinct salmonella clones from farm anima l s medicated with antibiotics i n subtherapeutic concentrations , through food products , to humans , who thus acquire cl inical salmonel losis . For example , a multiple ant ibiot ic-res istant strain of ï¿½ newport originated in farm anima l s exposed to chloramphenicol , a drug not approved for feed additive use , rather than to penicil l in or the tetracycl ines . We bel ieve further that appl icat ion of ava i l able methods for clonal analys is o f bacterial isol ates in future outbreaks of salmonel losis can provide the direct evidence required to rel ate use of spec i fic antimicrobials on the farm to human infect ion with antimicrobial -res istant foodborne pathogens ; or could show that no relationship exists . This model est imates risks for salmonel losis , because adequate data were l acking to quant i fy the risk for other foodborne pathogens , such as campylobacter jeiuni , Yers inia enterocol itica , and enterohemorrhagic ï¿½ ï¿½ . Furthermore , this report deals only with the risk for mortal ity from salmone l l osis , and does not estimate risk for morbidity due to lack of data . In contrast to the paucity of direct evidence impl icating subtherapeutic use of ant imicrob ials as a potential human health hazard , the committee found a cons iderable body of indirect or circumstantial evidence as fol lows : o Data on the biologic properties pf transposons and R plasmids . The use of ant imicrobial agents for either therapeut ic , prophyl actic , or growth-promoting purposes generates a strong selective pressure for the emergence o f drug-res istant bacteria in the environment , and thi s select ion operates a t severa l l evels s imultaneously by caus ing molecular expansion of the ant imicrobial -resistance genet ic elements as wel l . The basic genetic unit o f res istance , the transposon , can occupy almost any locat ion i n the genome , but most typ ical ly is o n a conjugat ive R plasmid . The ab i l ity o f the drug-res istance gene to transpose , as a transposon , and to move to other stra ins and spec ies provides the potential to confer res istance to the select ing ant imicrobial on a large population of bacteria . Stud ies of bacterial stra ins d i f fering only in the presence of an R plasmid have general ly revealed l ittle d i f ference in virulence . However , the results o f R plasmid acquis it ion can be striking when a conjugative R plasmid contains virulence genes such as ones encoding enterotoxin , aerobact in , hemolys in , or colonizat ion factors in addit ion to
3 antimicrobial-resistance determinants . Selection by antimicrobials can promote spread of virul ent stra ins in instances in which R plasmids have incorporated virul ence genes or in which virul ence plasmids have acquired drugÂ res istant transposons . Thus , the mere presence of the drugÂ res istant phenotype may be enough to enhance the diseaseÂ producing potential of the pathogen . Most spec ies of Enterobacteriaceae pathogenic for humans and anima l s carry at least one essential determinant of pathogenic ity on a plasmid . o There is evidence of extens ive use on farms and in feedl ots of subtherapeut ic concentrat ions of penic i l l in , the tetracycl ines , and other ant imicrobials ( see Chapter IV) . OVer 3 1 mil l ion pounds of antib iotics are produced annual ly in the United States . Although accurate data on ant ibiotic use in animal feeds are not ava i l abl e , estimates indicate that a lmost hal f the total annual production of antibiotics is directed to use in farm anima l s . The tetracycl ines used in l ivestock and poultry feeds represent almost 5 0 % of the total of antibacterial use in feeds . Almost 9 0 % of a l l ant ibiotics used i n f a rm anima l s and poultry is admini stered in subtherapeutic concentrations . About 7 0 % of the total of a l l ant ibiotics used in subtherapeutic concentrations in animal feeds is given for the purpose of disease prevent ion ( prophylaxis ) , and the remainder of this amount is administered for growth promotion . o Ample evidence exists of a high preval ence of ant imicrob ial res istance among isolates of salmone l l ae from farm anima l s . The frequency of resistance to any of the commonly tested antimicrobials among farm-animal isol ates of salmone l l ae ranges from 69 to 8 0 % ; of res istance to ampic i l l in , from 1 5 to 7 2 % ; and of res istance to tetracycl ine , from 3 7 to 8 1 % . These frequenc ies o f res istance among anima l isolates are 3 -5 t imes greater than those among stra ins isol ated from human be ings . Surveys of res istance to various antimicrobials among salmone l l a and ï¿½ co l i isolates from farm anima l s in the United States general ly have shown that feeding antibiotics in subtherapeutic concentrations increases res istance to antibiotics . The prevalence of res istance can vary cons iderably , probably with temporal and regional differences , and differences among farm animal spec ies . The variations may also re flect d i fferences in antibiotic practices , such as the use of speci f ic antibiotics and subtherapeutic vs . therapeutic dosages , the rel at ive proport ions o f res istance in bacterial isolates from each o f the f a rm animal spec ies , and env ironmental factors , such a s product ion methods , and stress o n the anima l s .
4 o Animal and poultry carcasses in meat-processing plants are o ften contaminated with intestina l pathogens and ï¿½ ï¿½ ( see recommendations in Chapter XI ) . Contaminat ion with salmonel lae has been found in up to 4 5 - 5 0 % of samples o f ready-to-cook poultry obtained from food markets a t d i fferent t imes and in d i fferent parts of the United States . Although few data are avai l able on the preval ence of ant imicrobial res istance among salmonel l ae isol ated from animal and poultry carcasses at meat-processing plants , it is l ikely that the prevalence would be s imilar among isolates from anima l s on the farm . The l imited data ava i l able on salmonel l a isolates from retail -marketed poultry would show about 3 0 -4 0 % are res istant to the tetracycl ines and about 6 5 % to one or more antimicrobials . o Human infection with salmonel lae or other enteric bacteria may fol l ow handl ing and ingestion of improperly cooked , packaged , frozen , or refrigerated meat or poultry contaminated with these organisms . o The results of several studies have shown the selection of drug-resi stance in col i form bacteria due to the use of antibiot ics in feed and subsequent spread from farm anima l s to humans . The potential exposure of members o f a farm family to various enteric bacteria indigenous to farm an ima l s has been investigated to determine the temporal frequency with which antibiot ic-res istant ï¿½ QQli stra ins from such contact spreads to them . Fol lowing the use o f tetracycl ine-supplemented feed in flocks o f chickens , the intestinal col i form flora in these chickens became largely tetracycl ine-resistant : within 2 weeks 9 0 % of the col i form isolates , predominantly ï¿½ QQli, from the chickens were tetracycl ine-resistant . The prevalence o f tetracycl ineÂ res istant col i form organisms also increased in the intest inal tracts of the 11 members of the farm family caring for the ch ickens , but not in members of neighboring fami l ies . After 5 - 6 months of use of subtherapeut ic concentrat ions of the tetracycl ines in chicken feed , 3 1 % of fecal samples taken at weekly interval s from members of the family conta ined over 8 0 % tetracycl ine-resistant col i form bacteria , in contrast to 7 % of the samples from neighbors . Multiple ant imicrob ial res istance , most l ikely plasmid-mediated , to unselecting ant imicrob ial s ( streptomycin , ampici l l in , sul fonamides ) , and to the tetracycl ines , developed in over 5 0 % of the ï¿½ Â£Qli stra ins isolated from ch ickens fed tetracycl ine-containing feed for over 10 weeks . Another study , described in Chapter VI , indicating that antibiotic-res istant ï¿½ col i of farm origin can spread to humans involved the use in pigs of a nonabsorbable antib iotic of the streptothricin group , nourseothricin , which had not been used in humans and therefore could not have spread initially from humans to farm
5 anima l s . After 2 years of use of this antibiotic in p iq feed , ï¿½ QQli that contained plasmids bearinq nourseothricin res istance were present in the feces of 3 3 % of piqs with diarrheal i l l ness , 1 8 % of workers on the p iq farms , 1 7 % of family members of the farm workers , and 1 6 % of outpat ients from the same qeoqraphic reqion where nourseothricin had not been used in feed . Even thouqh no nourseothricin had been used in treatment of humans in the reqion , 1% of the ï¿½ Â£Qli urinary tract infections o f outpatients were due to nourseothricin-res istant stra ins . In contrast , intestinal ï¿½ QQli isolated from outpatients in neiqhborinq reqions , where nourseothricin had not been used in piq feed , were not nourseothricin-resistant . Althouqh much of the important information needed to evaluate these observations is lackinq , the avai lable informat ion does appear to suqqest nourseothricin-resistant ï¿½ col i from p iqs were transmitted to humans and probably from humans to other humans . Whether nourseothricin-res istant isolates are more or less virulent than susceptible ones is not known . Epidemioloqic approaches have been used to address further the quest ion of whether exposure o f human beinqs to antibiot ic-resistant bacteria of farm-animal oriqin enhances the risk of subsequent infection with such stra ins . In one such study ( see Chapter VI ) , the possible acquisit ion of ï¿½ QQ1i urinary tract infections due to antibiot ic-resistant stra ins of farm oriqin was examined in a population of about 100 female employees in poultry process inq plants . ï¿½ ï¿½ were isolated in 9 5 % of the cultures from poultry : 9 6 % of these strains were res istant to one or more ant imicrobials and 87% were res istant to two or more druqs . Eleven percent of the female workers who had not recently taken antimicrob ial therapy had bacteriuria , with ï¿½ QQli account inq for two-thirds of the isolates . Of the l atter , 1 7 % were res istant to one or more ant ibiotics . The antibioqrams of the ï¿½ QQli from the bacteriuria workers and from the poultry to which they were heavily exposed seldom were s imilar or identical . The ï¿½ QQli stra ins from both sources were examined for commonly occurrinq R plasmids with restrict ion-endonuclease-diqestion patterns to trace spread of druq res istance from poultry to workers . The presence of a unique plasmid pattern endemic in poultry also found in the human isolates miqht support the concept of such spread . However , because the same plasmid could be demonstrated in two isolates in only a few instances , the results had too l ittl e power to exclude the possibil ity of spread . Unfortunately , intest ina l ï¿½ col i from the workers were not studied , and , there fore , poss ible spread to the ir qastrointest inal tracts could not be ascerta ined . The val id ity of us inq urinary tract infect ion as an epidemioloqic approach to the quest ion of whether exposure of humans to
6 antimicrobial -resi stant bacteria o f farm-animal oriqin enhances the risk of subsequent infection by such stra ins can be seriously questioned . ï¿½ ï¿½ strains caus inq urinary tract infections appear to represent a selected qroup of clones not included amonq those equa l ly l ikely to col onize anima l s and humans . Monitorinq of infections amonq farm workers and their famil ies caused by antimicrobial-res istant salmonel lae or other qastrointestinal pathoqens of farm animal oriqin miqht provide evidence of spread of these orqanisms to humans with ensuinq i l lness . Extens ive data bearinq on this issue are not ava i lable . In one study , salmonellosis or diarrheal i l lness did not occur at a hiqher frequency in farm chi ldren reqularly exposed to poultry than in a control qroup l ackinq such exposure . Althouqh scattered cl inica l case reports have documented transmiss ion of salmonellae from farm an ima l s to humans , this does not appear to occur commonly , or at least is not often recoqn i z ed . I n the United Kinqdom , s ince about 19 7 0 a l l antimicrob ials used in humans have been banned for use i n anima l s , thus only selected ant imicrob ials can b e used for animal qrowth promotion . The Swann Committee report of 1 9 6 9 made this recommendation and prohibited penici l l in , tetracycl ine , and certain antimicrobials from beinq used as feed additives . However , these druqs could be used for therapy or prophylaxis of disease only when prescribed by a veterinarian and then only for a l imited time period . Alternatively , antimicrob ials such as Zinc Bacitracin , Virqiniamycin , and Avoparcin have been used as feed addit ives but not for therapeutic indicat ions . Use o f these druqs does not select for bacterial stra ins res istant to penici l l in , tetracycl ine , or other antimicrobials used in human medicine . Some invest iqators in the United Kinqdom ins ist that the selection of res istant salmone l l ae ( and other bacterial stra ins ) is due to the therapeutic use of antimicrobial druqs in humans as wel l as animal s . The concentration of 2 0 0 qramsjton i s a therapeut ic or prophylactic dose in the United Kinqdom and must be prescribed by a veterinarian . In contrast , th is concentration in the USA is cons idered a subtherapeutic dose when it is administered for 2 weeks or l onqer and does not require a prescript ion from a veterinarian . Consequently , comparison of the ef fects of this concentration of antimicrobials on the select ion of res istant enteric pathoqens , especia l ly Salmonel l a , is d i f f icult because of the d i f ference in appl icat ions in these two countries . Althouqh penicil l in and the tetracycl ines have been banned as feed additives in the U . K . , both can be used when prescribed for therapy or prophylaxis of disease . The amounts of antimicrob ials used in veterinary practice has increased s ince the Swann Report . The surveys conducted subsequently indicate a fa irly constant and l ow incidence of
7 res istance amonq salmonella isolates except for ï¿½ typhimurium phaqe type 204C, an exceptional stra in that has become the most commonly isolated strain from cattle , especially calves . Not only bas it been reported in increas inq numbers s ince its isolat ion in 19 7 9 , but each year new ant imicrobials have been added to the l ist to which it i s res istant . In a few human cases this stra in bas developed serious treatment problems because of the l imited choice of effect ive antimicrobial s . Althouqh in 1985 it was respons ible for 4% of a l l reported cases of salmonel losis in humans , most of them were sel f-l imitinq . It i s d i f f icult to assess the ef fect of bann inq penicil l in and the tetracycl ines as feed additives in the U . K . a fter 1 9 6 9 when the Swann Report recommendations were implemented . There were no systematic col l ections o f stra ins prior to the report to serve as a basel ine . Also , there have been s iqni f icant cbanqes in the methods of ra is inq and market inq anima l s and these have been important factors in the spread of res istant Salmonella , e . q . , ï¿½ typhimurium phaqe type 204C. The multiple antimicrob ial res istance profile of this particular stra in presumably bas been caused by the many ant imicrobials used therapeutica l ly for scours in calves and not by subtherapeutic doses used in the U . K . Other stra ins o f Salmonel la have not shown this tendency to acquire res i stance plasmids under s imilar circumstances . Various other salmone l l a species , of animal oriqin , that cause human disease have fluctuated widely in incidence durinq this same period but there has not been any dramat ic increase in res istance to antimicrob ials in bacterial isolates . Some experts in the U . K . are convinced the Swann Committee recommendations have had no impact in reducinq one hazard this Committee was charqed to address , the selection of res istant bacteria by antimicrobials in animal feed . Because the committee was unable to find data directly impl icatinq the subtherapeut ic use of feed antimicrobials in human i l lness and that much of the ava ilable evidence was primari ly circumstantial , often amb iquous , and somet imes confl ictinq , the committee proceeded to devel op a risk model and perform a quantitative risk assessment . Even thouqh salmone l l os i s may contribute only a sma l l portion of the total incidence o f possible human di sease due to subtherapeut ic use of ant ibiotics in animal feed , our model was restricted to salmonel losis , because adequate data were lackinq to quanti fy the risk for other foodborne pathoqens , such as campylobacter jei uni , Yersinia enterocolitica , and enterohemorrbaqic ï¿½ coli. The model cons ists of a sequence of f ive mathematical ly derived quantitative estimates that are l inked in a cascade fashion :
s (1) Annual number of cases o f salmonellosis reported in the u.s. ( the committee has used a mid-range estimate of 50, 000 cases in developing the model in Chapter VII) (2) Fraction of salmonel la strains from human cases showing res istance to-- ( a ) any antimicrobial ( mid-range estimate , 2 4t ) ( b ) penicil l in/ ampicill in and/or tetracycl ine ( mid-range estimate , 1 St ) (3) Death rate associated with infection b y salmonella strains with various res istance patterns-- ( a ) susceptible to a l l antimicrob ials (mid-range estimate , O . St ) ( b ) res istant t o any antimicrobial (mid-range estimate , l . Ot ) ( c ) res istant t o penic i l l in/ ampici l l in andjor tetracycl ine ( mid-range estimate , l . Ot ) (4) Fract ion of deaths due to salmonel losis that are associated with salmonella stra ins of farm origin (mid-range estimate , 7 0 t ) (5) Proport ion of the above fraction ( 4 ) resulting from the subtherapeutic use of antimicrobial s in animal feed-- ( a ) any antimicrob ial ( mid-range est imate , SSt ) ( b ) penici l l in/ ampic i l l in and/or tetracycl ine ( midÂ range estimate , 90 t In view of the nature of the l ink between individual estimates described above , the five estimates can be multipl ied to indicate the number of annual deaths . With appropriate modification of the relevant estimates , the number of deaths due to use of antimicrobials in feed only for the purpose of growth promotion , rather than for a l l subtherapeutic uses ( both growth promotion and disease prevention ) , can be calculated . It is extremely important to recognize that the numbers of annual deaths est imated by mult iplying the f ive individual estimates l isted above are not necessarily excess deaths . It is possible that , even i f all subtherapeutic uses o f antibiotics were stopped , a l ike number of deaths might replace those produced by res istant strains as a result o f infections by drug-susceptibl e salmone l l ae .
9 It i s pos s ible to estimate the number of excess deaths due to subtherapeutic uses of antibiot ics by introducinq into the risk model the so-ca l l ed " etioloqic fraction . " This fraction represents the proportion of cases that would almost certainly not occur in the absence of druq-res istant stra ins . These excess cases are estimated by takinq into account the proportion of the population that is takinq antibiotics at any qiven time and the documented excess risk of infection fol l owinq such ant ib iotic administrat ion . Estimates based on inclusion of the " et iol oqic fract ion" are more certa in than those estimated without its inclus ion in the sense that these represent true excess cases . Because the data avai l able for use in the risk model are scanty , were l ikely to have been col l ected for other purposes and retrospect ively , and sometimes require extens ive extrapol ation , the inherent l imitat ions of the model must be appreciated . Furthermore , it should be recoqn i z ed that the fiqures used in the model result in a l ow estimate of the actual number of deaths , because poss ible unreported deaths from salmonellosis are not estimated and deaths due to other foodborne intest inal pathoqens , such as Campylobacter i eiuni , have not been taken into account . Moreover , there is no doubt an additional burden of disease ( morb idity ) should be considered part of the characterization of risk . Insufficient data were available to a l l ow quant i ficat ion of this burden . The committee bel ieves that it has empl oyed the best ava i l able data to make a series of l ow , mid-ranqe , and hiqh estimates for each sequential step in the model . A s imilar approach is taken to derive three est imates of the etioloqic fract ion . Because values for five variables ( described above ) have been multipl ied to provide specific risk quant i f ication , and because there are three different estimates of each variable ( low , mid , and hiqh ) , there are 243 different possible estimates of risk for each l inked sequence of steps . The committee tends to place qreatest rel iance on estimates near the mid-ranqe ( median ) value as be inq the " l ikel iest" est imate reported below . With this risk model , a series of est imates of annual numbers of deaths from salmonellosis attributable to subtherapeut ic uses of antimicrobials in animal feed were made : o The l ikel iest estimate of deaths attributabl e to subtherapeut ic uses of penicil l in and/or the tetracycl ines for both prophylaxis and qrowth promotion is in the ranqe of 40 per year . o The l ikel iest estimate of deaths attributable to subtherapeut ic uses of penicil l in and/or the tetracycl ines for qrowth promotion is in the ranqe of 15 per year .
10 It must be emphasi zed that neither of the above estimates is certa in to represent the true excess number of deaths due to subtherapeutic uses o f antibiotics in animal feed . It i s poss ible that stopping such uses wil l not reduce these numbers . Inclusion of the etiologic fracti on a l l ows estimati on of the excess numbers of annual deaths : o The l ikel iest estimate o f excess deaths attributable to subtherapeutic uses of penic i l l in and/or the tetracycl ines for both prophylaxis and growth promoti on is in the range of 6 per year . o The l ikel iest estimate of the excess deaths attributable to subtherapeutic uses of penic il l in and/or the tetracycl ines only for growth promotion only i s in the range of 2 per year . It i s also poss ible to estimate the numbers of deaths due to " increased di fficulty of treatment . " These estimates, which are more uncertain than the others , represent cases due to poss ibly increased virulence of res istant strains, presence of resistance to antimicrobials ord inarily employed in treatment of such infect ions when they are severe or occur in particularly vulnerable persons , some other factor, or a comb ination of these . These estimates probably include those in the etiologic fraction estimated above . o The l ikel iest estimate of deaths attributable to subtherapeut ic uses of penici l l in and/or the tetracycl ines for both prophylaxis and growth promotion and aris ing because of " increased diff iculty of disease treatment" is 20 per year . o The l ikel iest est imate of deaths attributable to subtherapeut ic uses of penici l l in and/or the tetracycl ines for growth promotion only and aris ing because of increased difficulty of disease treatment is 8 per year . The currently 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 speci f ic points of direct interest . Complete evaluat ion of these risk estimates o f mortal ity from salmonellosis attributable to subtherapeutic uses of ant imicrobials in animal feed requires cons ideration of the pos s ible benef its to food product ion that might accrue from such antimicrobial use . Cons ideration must also be given to the quest ion of whether overal l human deaths from salmonel los is ( attributable to both antimicrobial-
11 susceptible and -resistant stra ins ) would be changed by the discontinuance of subtherapeut ic use of penici l l in andjor tetracycl ine . The committee bel ieves that , although some deaths attributable to antibiotic-res istant stra ins may be substituted for by deaths from susceptibl e stra ins , the total number of deaths would decrease . However , thi s cannot be proved at present . Using al l the resources noted above , the committee was unable to find a substantial body of direct evidence that establ ished the existence of a definite human health ha zard in the use of subtherapeut ic concentrations of penici l l in and the tetracycl ines in animal feeds . The committee does not offer recommendations for risk management or pol icy making , because thi s was not part of its mandate . However , a series of recommendat ions for strengthening the data bases for future risk analyses have been made by the committee . Many of these would warrant implementation in order to monitor antib iotic use and the changes that might ensue in druq res istance in isolates from anima l s and humans with salmonellosis and other foodborne diseases . These recommendations would seem particularly appropriate in view of the fact that debate on the benefits of use of subtherapeutic doses of penici l l in and the tetracycl ines in animal feed has qone on for over two decades .
II INTRODUCTION Subtherapeutic concentrations of antimicrobials in feed have been used for decades in the rai sing of anima l s for food production . The concentrations are lower than those usua l ly chosen to treat estab l i shed infections in animals , but high enough to a ffect growth of bacterial components of the gastrointestinal flora . Subtherapeutic concentrations are . used to improve growth and to prevent infection during periods of increased suscept ibil ity in rearing . In 19 8 0 , a committee in the Divis ion of Medical Sciences of the Nat ional Res e arch Counc i l pre pared a report entit led The Ef fects o n Human H ea l th o f Subtherapeutic Use o f Ant imicrobi a ls in An ima l Feeds2 . The comm ittee concluded that h a za rd s to h uman heal th a s s o c i ated w i th subthera peut ic use o f ant imicr o bia ls in animal feeds had been neither proved nor disproved ; that was not to say that the postulated hazards did not exist . That committee made a number of observat ions that are st i l l relevant : o Subtherapeutic uses of antimicrobials in anima l s increase the preval ence of antimicrobial res istance i n some bacteria , such as salmonel lae and Escherichia Â£211 . Persons in close contact with animals treated with ant imicrobial s are more l ikely to harbor antimicrobial -res istant ï¿½. QQli than are persons not so exposed . However , in studies of antimicrobial -resistant ï¿½. QQli in humans with animal contact , the dosage and durat ion of antimicrob ial use in the animals have not been clearly defined . Thus , subtherapeut ic use usual ly cannot be distinguished from therapeutic use . o S laughterhouse workers carry some of the same phage types of Enterobacteriaceae as are found in sl aughtered anima l s and in the slaughterhouse environment . However , the relevant studies were not conclus ive , because too few persons were examined . In addit ion , the anima l s to which the workers had been exposed had probably received both therapeutic and subtherapeutic doses of antimicrobial s . o A l ink could not be establ ished between il lness due to ant imicrobial -res istant pathogenic bacteria and contact with anima l s given only subtherapeutic antimicrobials or ingestion of meat from such animals . 12