National Academies Press: OpenBook
« Previous: THE RISK MODEL: OVERVIEW OF THE PROBLEM AND NEED FOR A MODEL
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 149
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 150
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 151
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 152
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 153
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 154
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 155
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 156
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 157
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 158
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 159
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 160
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 161
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 162
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 163
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 164
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 165
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 166
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 167
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 168
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 169
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 170
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 171
Suggested Citation:"THE ESTIMATION OF RISK." Institute of Medicine. 1989. Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed. Washington, DC: The National Academies Press. doi: 10.17226/19030.
×
Page 172

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.

150 ranges . We caution that such numerical estimates are incomplete descriptions of risk and should not be used without citation of the associated uncerta inties , many o f which can not be expressed quant itatively . METHOD OF RISK CALCULATION Chapter VI I expla ined the risk model , provided a view o f the uncertainties associated with each parameters used by the committee and presented the committee ' s l ow , mid-range , and h igh estimates of these parameters required by the model . The estimates of risk presented in this chapter are l im ited to infections with ant ibiotic-res istant stra ins ( due to subtherapeutic uses of ant ib iotics in anima l s ) of Salmone l l a ( other infectious bacteria have been mentioned , but risks are not calculated for them ) and further l imited to annual numbers o f deaths from these infections ( risk due to morbidity and from non-lethal cases were not calculated ) . Annual numbers of deaths are estimated by a stra ight­ forward multipl ication of combinations of parameter est imates , as il lustrated for the mid-range estimates alone at the opening of Chapter VI I . The committee could not select a s ingle "best estimate" of any of the parameters for use in describing risk . Al l poss ible combinat ions of the parameters were thus used to produce a range of poss ibl e risks . The various combinations sel ected for estimating d i f ferent risks are described in the text to fol l ow . I n each case the complete set of possible ri sks for each combination was calculated , and the results were converted to percent iles and pl otted on graphs as cumulat ive distributions ( Figures VI I I - 1 through VI I I - 12 ) . The committee attaches some importance to the 5th , 5 0th ( median ) and 9 5th percentiles as descriptions of the whol e set of estimates . However , the 5th and 9 5th percent i l es are not to be interpreted as confidence l imits , because they do not reflect any underlying probab i l ity distribution . A given percentile s imply describes the fraction of risk est imates that fal l bel ow it . Thus , for example , if a spec i f ic risk i s based o n multipl ication of five d i f ferent parameters , and there are three d i f ferent estimates of each parameter ( l ow , mid , and high-range ) then there are 2 4 3 d i f ferent possib l e � estimates of r i s k ( 3 ) . The l owest five percent of these 2 4 3 estimates fal l s bel ow the 5th percent i l e , etc . The committee bel ieves that it is unl ikely that all ( or nearly a l l ) of the l ow or high estimates hold s imultaneously . Thus , combinations of parameters that are all ( or nearly a l l ) low , or high , are highly implaus ibl e . This i s based , i n part , on the improbab i l ity of being cons i stently wrong in the same direction , and in part on the committee ' s attempt to set the l ow and high estimates at the bounds of general plausib i l ity .

1005\t � � 70% IJ u 600; � �IJ 50% Ja:1 'a 40% � ti; t-J 30% 20% 1 0% � 0 200 400 600 800 Eatilna.t.ed death/year Figure VI I I - 1 . Estimates o f annual numbers o f deaths from subtherapeutic uses o f any antibiotic for both prophylaxis and growth promotion ( multipl ication of l ines 1 , 2 a , 3 b , 4 , s a o f Table VI I I - 1 ) .

�� � � 70. • u 60Sll � j • 5o. fl1 � 40Sll • ...... U'l "' 30Sll 20. 10. 0. 0 1 00 200 300 400 Eatbna.ted death/year Figure VI I I - 2 . Estimates o f annual numbers o f deaths from subtherapeutic uses o f pen ic i l l in/tetracycl ine for both prophylaxis and growth promot ion ( multipl icat ion of l ines 1 , 2 b , 3 c , 4 , 6 a of T ab l e VI I I - 1 ) .

�� 90Sll 8<* 7� • u � � � • � � '8 "'* • ..... 1.11 w 3()Sl; 2� 1� � 0 1 00 200 300 400 Eatbna.ted death/year Figure VI I I - 3 . Estimates of annual numbers of deaths from subtherapeut ic uses of any antibiotic for growth promot ion only (multipl ication of l ines 1 , 2 a , 3 b , 4 , s b of Table VI I I - 1 ) .

�� � 80S\; 7� • u � � �• 50% J!;l -a 40S\I � ..... 1.11 ol:oo � 2� 1� � 0 40 80 120 160 200 240 280 320 Estimated death/year Figure VI I I - 4 . Estimates of annual numbers o f deaths from subtherapeutic uses of penici l l injtetracycl ine for growth promot ion only ( multipl ication of l ines 1 , 2 b , 3 ° , 4 , 6 b o f T ab l e VI I I - 1 ) .

100. � � 7<* • u 6<* � j • 5<* � "8 � � � U1 3<* 2<Mll 1� <* 0 40 80 120 160 200 Eatbnated death/year Figure VII I - 5 . Est imates of annual numbers of deaths in the etiologic fraction attributable to subtherapeutic use o f any antibiotic for both prophylaxis and growth promotion (multipl ication o f l ines 1 , 7 , 3 b , 4 , s a of Table VI I I - 1 ) .

100S\I 90\ll � 7<* ID 6<* u +I � s - 5<* +I ID IJ;1 � 4<* ..... U1 0\ 30� 2<* 1<* <* 0 4-0 80 1 20 160 200 Eatiln.ated death/year Figure VI I I - 6 . Est imates of annual numbers of deaths in the etiologic fraction attributable to subtherapeutic uses of penicill in /tetracycl ine for both prophylaxis and growth promot ion ( multip l icatio n o f l ines 1 , 7 , J c , 4 , 6 a of Table VI I I - 1 ) .

100� � 80� 70� l'l v .,J 60Sli d a ... .,J 50Sli Ill rz1 'M 0 40Sli � .... U1 -..J 30� 20� l<>sli OSli 0 20 40 60 80 100 1 20 Eetilnated death/year Figure VI I I -7 . Estimates of annual numbers of deaths in the etiologic fraction attributable to subtherapeutic uses of any antibiotic for growth promotion only (multipl ication of l ines 1 , 7 , 3 b , 4 , s b of Table VI I I - 1 ) .

�� 90ll � 7� Ill u 60� +I Gl � +I 50Sli II J-:1 '0 40loll � .... U1 co 3� 2� ·� � 0 20 40 60 80 100 1 20 140 Eetim.ated death/year Figure VI I I -8 . E st ima t e s o f annual numbers o f deaths in the et iologic fract ion attributabl e to subtherapeut ic uses of enic i l l injtetracycl ine for growth p romotio n only g (multipl ication of l i n es 1 , 7 , 3 ° , 4 , 6 of Tabl e VI I I - 1 ) .

100. 9<* eo. 70. • u 6o. � � • so. � '0 4o. � � \0 ao. 2o. 10. 0. 0 20 40 60 80 1 00 120 140 160 1 80 Estbna.ted death/year Figure VI II-9 . Estimates of annual numbers of deaths aris ing because of higher death rate and increased d i f ficulty of disease treatment attributable to subtherapeut ic uses of any ant ibiotic for both prophylaxis and growth promot ion (multipl icat ion of t imes 1 , 2 a , ( 3 b - 3 a ) , 4 , sa of Table VI I I - 1 ) .

1� � ao. 7� • u 60li � j • � � � � � ...... 0\ 0 3� 2� 10S\i � 0 20 .w 60 80 100 1 20 Eeti1118.te d death/year Figure VI I I - 1 0 . Estimates of annual numbers o f deaths aris ing because o f increased difficulty of disease treatment attributable to subtherapeutic uses o f penic i l l in/tetracycl ine for both prophylaxis and growth promotion (multipl ication of l ines 1 , 2 b , ( 3 c - 3 a ) , 4 , 6 a o f Ta b l e VI I I -1 ) .

1� � SOli 7� • u 60li � �II � r-:1 1M 0 ·� � ..... 0\ ..... 3� 2� ·� � 0 20 40 60 80 Eetiln.ated death/year Figure VI I I - 1 1 . Estimates of annual members of deaths aris ing because of increased d i f ficulty o f disease treatment attributable to subthera geutic uses of any ant ibiotic for growth promotion only ( multipl ication of l ines 1 , 2 a , ( 3 - J a ) , 4 , s b of Table VI I I - 1 ) .

1� � � 70ll • &I 60ll .., II �• SOli Ja:1 'a 40ll � ..... 0'\ t..) 30Sll 20ll lOll Oil 0 20 40 60 80 Eathnated death/year Figure VI I I - 1 2 . Estimates of annual numbers of deaths aris ing because of increased d i f f iculty o f disease treatment attributable to subtherapeutic uses of peni c i l l i n; t et racyc l ine f o r growth promot ion o n l y ( mult ipl icat ion o f l ines 1 , 2 b , ( J c - J a ) , 4 , 6 b o f Ta b l e VI I I - 1 ) .

163 The committee thus tends t o place greatest rel iance on estimates near the median value . ESTIMATION OF RISKS Chapter VI I , " The Risk Model " , provides the data on which the committee has based its estimates o f human health r i sk that may be assoc iated with the subtherapeut ic use o f antibiot ics in animal feed . Table VI I I - 1 gives the parameter estimates used in assess ing the whole o f the problem of exces s human salmone l l o s i s deaths that might be attributable to any low-level farm use of ant ib i ot ics . These estimates are taken d i rectly from the text and tables in Chapter VI I . Var ious combinat ion ' s o f these parameters are used to est imate d i fferent types of risk . Twelve d i fferent sets of r i sk est imates were produced from these data : these est imates are presented graph ical ly in F igures VI I I - 1 through VI I I - 1 2 , and are summari zed in Table VI I I -2 . The bas i s and meaning o f the twelve d i f ferent sets o f est imates that are shown in F igures VI I I - 1 through VI I I - 1 2 are described in the fol lowing text . MAXIMUM POS S I BLE NUMBERS OF EXCESS DEATHS F igure VI I I - 1 shows the cumulat ive d i stribut ion o f the 2 4 3 est imates o f annual deaths generated for drug res i stance of any type { Table VI I I - 1 , multipl icat ion of parameters from l ines 1 , 2 a , 3 b , 4 , s a ) . F igure VI I I -2 shows the corresponding d i stributions for res istance to penic i l l in andj or tetracycl ine spec i f ical ly { Table VI I I - 1 , mult ipl icat ion o f parameters from l ines 1 , 2 b , 3 c , 4 , 6 a ) . Est imates presented in Figures VI I I - 1 and VI I I -2 represent subtherapeut ic use of ant ib iot ics for both growth promot ion and prophylaxis . F igures VI I I - 3 and VI I I -4 are s imilar to F igures VI I I - 1 and VI I I - 2 except that the est imates are for farm use o f ant ibiotics for growth promot ion only , rather than for a l l subtherapeutic uses o n the farm . F igure VI I I - 3 descr ibes risks for use of any ant ibiotic and is based on mult ipl icat ion of l ines 1 , 2 a , 3 b , 4 , and s b of Table VI I I - 1 . F igure VI I I -4 descr ibes r isks for use of pen i c i l l in and tetracycl ine only and is based mult ipl icat ion of parameters from l ines 1 , 2 b , 3 c , 4 , and 6 b in Table VI I I - 1 . Readers are caut ioned that these are not necessarily " excess deaths " in the sense that the total number i s increased by the quant ity ind icated : they are rather est imates of the annual numbers of deaths attributable to salmonel los is of the spec i f ied origin . These may , to some extent , overlap or repl ace deaths { in these pat ients or

164 TABLE VI I I - 1 DATA USED TO ESTIMATE ANNUAL DEATHS FROM SUBTHERAPEUTI C USES O F ANTI BIOTICS IN ANIMAL FEED Estimates Low Mid-Banqe High 1) Reported S almone l l a per year 4 0 , 000 5 0 , 000 65 , 000 2) Re s i stance o f Salmone l l a to ant imicrobials a) Res istance to any antibiotic *0 . 16 0 . 24 0. 31 b) Res istance to penicil l in/ 0 . 10 0 . 15 0 . 20 tetracycl ine 3 ) Death Rate -- infection by strains a ) Not res istant to any antibiotic 0 . 002 0 . 005 0 . 01 * *b ) Res istant to any antibiotic 0 . 002 0 . 01 0 . 04 * *c ) Resistant to penici l l in/ 0 . 002 0 . 01 0 . 04 tetracycl ine 4 ) Fraction of those deaths 0.5 0.7 1.0 associated with strains of farm origin 5 ) Fraction of antibiotic res istance of farm origin caused by subtherapeutic use of any antib iotic in animal feed a ) Prophylaxis and growth promot ion 0 . 85 0 . 88 0 . 92 b) Growth promotion only 0 . 05 0 . 25 0 . 50 6) Fract ion of antibiot ic res istance caused by subtherapeut ic use of penic i l l in/tetracycl ine in animal feed a) Prophylaxis and growth promotion 0 . 86 0 . 90 0 . 94 b ) Growth promotion only 0 . 05 0 . 30 0 . 60 7 ) Etiologic Fract ion 0 . 005 0 . 02 0 . 09 Source : Table prepared by the committee . The bases for a l l estimates are provided i n Chapter VI I . Note : p e n icil l i n/ tet racycl i n e = p e n i c i l l in , amp icil l i n or t e tra cycl ines . * Deci ma l fract i on or pr opo rt i on . ** The c omm i tt e e could not find evidence su f f i c i ent to j ust i fy the use o f d i f ferent death rates for stra ins res istant to one drug rather than another , or for multire s i stant stra ins vs . strains res istant to only one drug .

165 TABLE VI I I -2 SUMMARY OF THE VARIOUS ESTIMATES OF ANNUAL NUMBERS OF DEATHS FROM SUBTHERAPEUTI C USES OF ANTIBIOTICS Sgyrce of Etti11te1 lett Ettllltt' � <meslien> F i gure ¥1 1 1 · 1: 70 5 · 700 F i gure ¥1 1 1 ·2 40 1 · 400 F i gure ¥1 1 1 ·3 ; 20 1 · 400 F i gure ¥1 1 1 ·4 15 1 · 300 F i gure V l l l ·5� 06 1 - 200 F i gure Vl l l -6 06 1 · 200 F i gure VI I I · 7f 02 0 · 1 00 F i gure Yi l t -aJ 02 0 · 1 00 F i gure Vl l l ·9k 40 3 · 200 F i gure Vl l l · 10l 20 2 · 1 00 F i gure Vl l l · 1 1• oa 1 · 1 00 F i gure vr r r - 1 2 n oa 1 · 1 00 source : Adapt ed by t h e com. i t t ee f r� T eb l e V I J I - 1 end F i gure! ¥ 1 1 1 · 1 th rough ¥ 1 1 1 · 1 2 . e SOl of ett i ., t et f e l l be l ow t h i t f i gure ( rounded t o one t f gn i f i cent f i gu r e ) . b 51 of ett i ., tet f e l l be l ow l ower end of the renee; 951 of ett i .. t et f e l l be l ow upper end . E t t i ., t et of ennue l nu.btrt of dee t h t : c f ro. tubtherepeut i c utet of eny a s t i b i ot i c for both prophy l ex i t and g rowt h pro.ot f on ( .u l t i p l i ce t i on of l i net 1 , 2 , 3 , 4 , 5 8 of T eb l e ¥ 1 1 1 • 1 ) . a d f ro. !Ubt herapeut i c Ute! Of pen i c i l l i n/�e t r acyc l i ne for both prophy l tX i l end g rOWt h pro.ot f on ( .u l t i p l i ca t i on of l i net 1 , 2 , 3 , 4 , 68 of T eb l e V l l l - 1 ) . c 8 f ro. tubthere � t f c u&e• of eny ent i b f ot i c for g rowth pro.ot i on on l y ( .U l t i p l i ce t i on of l i ne! 1 , 2 8 , 3 , 4, 5 of T ab l e V I J J · 1 ) . f f ro. tubthertptUt i C Ute! Of esn i c i l l i n/ t st rtcyc l i nt for g rowth pro.ot i on on l y c ( .u l t i p l f ce t f on o f l f net 1 , 2 , 3 , 4 , 6 o f Teb l e ¥ 1 1 1 · 1 ) . I i n the et f o l og f c f r ec t f on e t t r i but eb l e to tubtherapeut f c ute of 'BY ent f b i o t i c for both prophy l ex i t end growth pro.ot f on ( .u l t i p l i cat f on of l i net 1 , 7 , 3 , 4, s • of T eb l e Y l l l - 1 ) . h f n the et i o l og i c f ract i on e t t r f but ebl e to tubtherepeut i c Ultl of pen i c i l l i n /tet recyc l i ne for both prophy l ex f t end g rowth pro.ot i on ( .u l t f pl f cet i on of l i nt! 1 , 7, 3 , 4 , 68 of T eb l e c ¥1 1 1 · 1 ) . i n the et f o l og f c f re c t l on e t t r f but eb l e to tubth B rapeut &c utet o f eny ent f bf ot f c f o r grow t h pro.ot f on on l y ( .u l t f p l f c at i on of l i ne• 1 , 7 , 3 , 4 , 5 o f T e b l e V l l l · 1 ) . f n t h e et f o l og f c f ract i on a t t r l bu t eb l e t o tubtherepeut i c utet of pen i c i l l i n/ t e t r ecyc l f ne f o r growth pro.ot i on on l y ( .u l t f p l i ce t f on of l i nt! 1 , 7 , 3 , 4 , 6 of T e b l e Y l l l • 1 ) . c b k e r i t f ng beceute of h i gh e r dee th rete end l nc reeted d i f f i cu l ty of d i teate t re e t •ent a t t r f but e b l e to tubtherepeut i c utea of eny a s t i bf ot i c for bo t h prophy l ex f a and growth pro.ot i on ( .U l t i p l f c e t f on of t f .,a 1 , 2 8 • ( 3 · 38 ) , 4, 5 8 of T tb l e V l l l · 1 ) . e r f a f ng because of f nc r eeatd d i f f i cu l ty o f d i aeeae t r ea t.,nt e t t r i bu t eb l e to aubt herepeu t l c uses of peG i c f l l i n/ t e t recyc l f ne for bot h prophy l ex f a end g rowth pro.ot f on ( .u l t f p l i c a t i on o f l f nea 1 , 2 , ( 3 - 3 ) , 4 , 68 o f T ab l e Y l l l - 1 ) . c a . • e r i a i ng beceuae of I nc reased d i f f i cu l t y of d f aeaae t reat.,nt e t t r l bu t ab l e t o a ubt he � apeut l c u&e• of eny ent i b i ot i c for g rowth promo t i on on l y ( .u l t f p l i c e t l on of l i nea 1 , 2 , ( 3 - e 3a ) , 4 , 5 of T eb l e V I I I - 1 ) . " E a t l .. tea o f annue l nu.btra o f dee tha e r l a i ng because ·of f nc r eeaed d i f f i cu l t y o f d i sease t rea t.,nt e t t r i bu t eb l e to a ubt h e r a peut i c use & of � n i c f'l l i n/ ,et recyc l i ne f o r g rowt h pro�t f on on l y ( .u l t f p l i ce t i on of l i nea 1 , 2 , (3 · 3 1 ) , 4 , 6 of T e b l e V I J I - 1 ) .

166 others ) from salmonellosis that would have occurred anyway with some other strain . Conversely , it is possible that these estimates underest imate the real number of excess deaths i f , for example , res istant strains tend to be more virulent than drug-susceptible stra ins , or i f the estimates in success ive l ines of Table VII I - 1 are not independent , ( see Chapter VI I regarding independence ) . ESTIMAtES OF DEAtH 8ASED ON ETIOLQGIC FRACTION As expla ined above , the est imates for deaths from all Salmonella strains with drug resistance attributable to l ow-level farm uses of antibiotics are not necessarily estimates o f the excess number of salmone l l os i s deaths from such use . A fraction of the excess can , however , be estimated--the " etiologic fraction" discussed in this sect ion and the death o f farm origin " harder-to-treat fraction di scuss ion in the fol l owing section . These two fractions may overlap ( e . g . , f igures for the etiologic fraction may ref lect some increase in the difficulty in providing effective treatment ) and , further , these two fractions do not necessarily account for the whole e f fect of farm use o f subtherapeutic leve l s of antibiotics ( e . g . there may b e a d i fference in virulence ) . Estimates for deaths attributable to the etiologic fraction--that is , cases o f salmonel losis that would s imply not have occurred in the absence o f res istance--requi re some modi f ication in approach . Parameter estimates are given in Table VII I - 1 . The odds ratios in Table VI I-7 are cal culated for the whol e population of exposed persons ; of these , some proport ion harbor res istant stra ins . The estimated odds ratios would be larger--perhaps substantially larger-- i f they were calculated to express the risk in persons who harbor such res istant stra ins . Use of the odds rat ios in Table V I I - 7 , therefore , a lready incorporate a reducat ion factor to express the risk in the population as a whole . Furthermore , thi s automatical ly reflects the actual proportion o f persons who have res istant stra ins ( perhaps in addition to susceptible stra ins ) and does not depend on the kind o f estimate in l ine 2 of Table VI I I - 1 , which dea l s with proport ions o f stra ins rather than with the whole set o f res istant stra ins that may inhab it one person . Thi s approach ignores the l ikel ihood that persons within fami l ies , within hosp ital wards , or otherwise in proximity may tend to carry the same stra ins of salmonel lae , but no data on thi s seems to be ava i l able for use here . Because of the frequency distribution of res i stant stra ins al ready incorporated into the odds ratios ( l ine 7 ) , no further adj ustment for res istance ( l ine 2 ) is needed or appropriate .

167 Perhaps future research studies can estimate odds ratios for the " etiologic fraction" of cases among persons who are hosts to one or more res istant stra ins . The odds ratios are l ikely to be substantial ly higher , but wi l l be reduced by the ( then appropriate ) inclus ion of such factors as those in l ine 2 . Unt i l thi s kind of additional information i s ava i l able , we bel ieve that our present analytic approach to the etiologic fraction is correct . In addition , the committee is concerned that death rates in the " etiologic fraction" : ( see l ine 3 ) may be above average , because some persons who rece ive antibiotics do so because of conditions related to immunosuppress i on , general debi l ity , or other i l l nesses that may damage normal body defenses . In the absence of data , however , the committee has chosen to apply the death rates in l ine 3 o f Table VII I - 1 to the " etiologic fract ion" . Two sets of estimates are presented for the " et iologic fraction" component of salmonel losis . F igure VI I I - 5 presents the 2 4 3 estimates for deaths in the etiologic fraction attributable to subtherapeutic farm use o f any ant ibiotic for both prophylaxis and growth promotion (mul t ip l ication of parameter in l ines 1 , 7 , 3 b , 4 , s a , of Table VI I I - 1 ) , and Figure VI I I - 6 presents s imilar estimates for penici l l in/amp i c i l l in andjor tetracycl ines uses only ( multip l ication of parameters in l ines 1 , 7 , 3 c , 4 , and s b o f Table VI I I - 1 ) . Figures VI I I -7 and VI I I -8 present s imi l ar estimates for growth promotion alone : Figure VI I I -7 concerns farm use of any antibiotic and Figure VI I I -8 concerns uses of pen ic i l l in/ampic i l l in and/or tetracycl ines only . EXCES S DEATHS DUE TO INCBEASED DIFFICULTY OF TREATMENT While few or no strains of sa lmonel lae are res istant to all cl inica l ly useful ant imicrobials in the modern therapeutic armamentarium , some individual drugs are potentially toxic , have unwanted e ffects in part icul ar groups of patients , may requi re parental administration , and some are very expens ive . Further , critical t ime i s required to determine patterns of res istance of bacterial isolates in spec i f i c infections . Thus , it would be medica l l y inappropriate , t o treat each suspected case of salmonel losis with the whole combination of antimicrobials that could conceivably be e f fective . More selective therapy is med ical ly appropriate , but it has the unfortunate e ffect in some cases of delaying or replacing treatment by the opt imum drug or drug comb inat ion , and as a result death rates may be higher in salmonellosis with res istant stra ins than with susceptible stra ins . Whatever the reason ( s ) , it has been commonly observed that infections with res istant stra ins of salmonellae more often end in death than infect ions with susceptible stra ins ,

168 suggested by l ines 3 a , 3 b , and 3 c of Table VI I I - 1 . The d i fference between these l ines can be interpreted as an index of the increased difficulty of providing effect ive therapy in cases of res istant salmonel losis . Because the est imates in l ines 3 a , and 3 b and 3 c are so cl osely l inked , the committee s imply worked with the three d i fferences ( at l ow , mid-range , and high level s ) rather than the 9 poss ible comb inations . Estimates of the s i z e of thi s effect for a l l subtherapeutic uses of any ant ibiotic are presented i n Figure VI I I -9 (mult ipl ication and parameters in l ines 1 , 2 a , ( 3 b - 3 a ) , 4 and s a , Table VI I I -1 ) and s imilarly in Figure VI I I - 1 0 for res istance to penic i l l in/ ampici l l in and/ or tetracycl ine antibiotics ( l ines 1 , 2 b ( 3 c -3 a ) , 4 , 6 a ) . S imilar figures , but l imited to drug use for growth promot ion , are given in F igure VI I I - 1 1 ( l ines 1 , 2 a , ( 3 b - 3 a ) , 4 s b ) for any drug res istance and F igure VI I I - 1 2 ( l ines 1 , 2 6 , ( 3 c -3 a ) , 4 , 6 b ) for peni c i l l in or tetracycl ine res istance . SUMMABY OF NQMERIC BESULTS Each o f the figures in this chapter presents a range of r isks , reported as annual numbers o f deaths . This procedure was used because the committee had no bas is for selecting any s ingle " best " est imate . This procedure produces , for the data in each f igure , a total of 2 4 3 estimates . The committee bel ieves that the best s ingle est imator is the median of the 2 4 3 estimates , and that the range from the 5th to 9 5th percentile is quite l ikely to contain the unknown true value . Because o f the way these estimates were devel oped they do not prov ide ordinary statistical confidence l imits ( as explained above ) , but they should in pract ice provide even greater certa inty than , say , 9 0 % or 9 5 % confidence l imits . The committee bel ieves that the minimum and maximum est imates presented in the f igures are not scient i f ical ly p l ausible because they would require that the mid-range estimates for the parameters ( Table VI I I - 1 ) a l l be cons istently or nearly consistently wrong by a large margin and all be in the same direction . Figures are presented to one decimal to emphas i ze that they are estimates , not counts . Data from the twelve figures are summa ri zed in Table VI I I -2 . The fol lowing is an i l lustration of how the f igures and data in Table VI I I - 1 are to be read : Figure VIII-1 . Est imates of annual numbers of deaths from subtherapeut ic uses of any ant ibiotic for both prophylaxis and growth promot ion . Figure VI I I - 1 is read as fol lows :

169 ( i) F ive percent of the estimates fal l below 5 to 6 deaths per year , and 9 5 % fal l below 7 0 0 deaths per yea r . Thus , the committee bel ieves that the true number is very l ikely to be between 5 and 7 0 0 deaths per year . ( i i ) The l ikel iest estimate is 7 0 deaths per year . Thi s i s the committee ' s best s ingle estimate for mortal ity in this category . The estimates in Table VI I I - 2 are derived from Figures VI I I - 1 through VI I I - 1 2 , and each range is based on d i fferent assumption ' s regarding uses ( e . g . , any ant ibiotic vs . penic i l l in/ampici l l in and , or tetracycl ines only , on prophylaxis and growth promotion uses vs . growth promotion only ) . The ranges a l so d i f fer with regard to other assumpt ions ( e . g . , inclus ion of " et iologic fract ion" , cons iderat ion of increased d i f f iculty of treatment ) . The spec i f ic meaning of each set of estimates i s indicated by the F igure headings , that are reproduce at the foot of thi s table . INTERPRETATION OF BESULTS The various estimates of risk presented in Table VII I -2 are based on somewhat d i fferent assumpt ions and have d i f ferent meanings , as indicated in the foregoing text and as summarized in the Figure headings � For each set of estimates the committee p laces greatest rel iance upon the 5 0th percent i l e figure , which has been termed the " l ike l iest est imate " in Table VI I I -2 . The range shown in Table VI I I - 2 almost certa inly encompasses the true figures . The committee i s not able to ass ign a numerical probabi l ity to the l ikel ihood that the estimates shown are correct . As noted ea r l i e r , none o f the se e s t ima t e s h a s been veri f i ed by trad i t i on a l s c ient i f ic methodo l og i e s ( i . e . , exper iment a l o r we l l - c ontro l l ed f i e l d stud i e s ) , and thu s shou l d be i nterpreted as s c i ent i f i c hypothes e s about the pos s ib l e extent o f the prob l em that a re cons i stent w i th a l l ava i l ab l e s c i e nt i f i c i n f o rmat i on . The comm ittee knows o f no d i rect ev idence t o support the se est imate s . They shou l d be con s i dered as hav i ng s c i ent i f i c support rough l y comp a rab l e to that ava i l able for est imates of l ow dose carcinogen ic risk associated with chemical carcinogens subj ect to regul at ion . The est imates of death presented in Table VI I I - 1 can be placed in the context of other types o f risk est imates . FDA , f or example , general ly holds that , for carc inogenic drugs used in anima l s that l eave toxic food res idues , l i fet ime risks of cancer ( presumed to be equ ivalent to l i fetime r isks of death ) , of around 1 0 -6 or less are of ins ign i f icant pub l i c

170 · health consequence . Using thi s yardstick , and assuming the enti re populat i on of the United States to be potentially exposed to such res idues , the numbers o f excess annual cancers ( i . e . , deaths ) , a ssuming the risk to be accurately known , can be estimated for one such drug as fol l ows : 2 4 0 x 1 0 6 persons x 1 0 -6 = 2 4 0 l i fetime deaths , or = 3 - 4 deaths per year . The total number of deaths due to carcinogenic res idues depends on the number of such drugs in use . Actual numbers of deaths are probably much l ower than these f igures indicate , because actual res idue level s rarely approach the maximum a l l owabl e and because it is unl ikely that most o f the populat ion i s exposed to these drugs on a cont inuing bas i s . Moreover , the risk estimat ion method used for carc inogens is des igned to overstate risk . That i s , the procedures used to est imate excess cancer risk include adoption o f upper 9 5 % con f idence l imits on the dose-response curve and several other assumpt ions about interspecies and high-to-l ow dose extrapolation that almost guarantee that the actual numbers of deaths w i l l be l ess than those shown above . In fact , actual risk may be zero . The above f igures are helpful nonetheless , because they reflect the hypothetical number o f excess deaths that might be cons idered of negl igible pub l i c health consequences . The estimates o f annual numbers o f excess deaths presented in Table VI I I - 1 are derived by a method that is not strictly comparable to that used by FDA for carcinogenic drug residues , so care must be taken in comparing these two sources of risk . However , no better bas i s for comparison is known to the committee , and , with the appropriate qua l i f icat ions , the drug-res idue cancer risks , apparently cons idered acceptable by FDA , do prov ide a moderately useful yardstick aga inst which the r isks in F igure VI I I - 1 can be measured . Moreover , the committee does not mean to suggest that the r isks cons idered acceptabl e by FDA for carc inogenic an imal drug res idues are necessarily appl icable to the determinat ion of acceptab i l ity of the risks that are the subj ect of this report . Whether the ri sks presented in Table VI I I - 1 are to be cons idered acceptable or unacceptable depends on many factors that fal l outs ide the scope of the committee ' s charge . Such determinat ions of acceptabi l ity are risk-management decis ions and thus are properly l e ft to FDA .

17 1 EFFECTS OF DISCQNTINQATI ON Wil l the number of deaths from salmonel losi s and its compl ications be reduced or otherwise altered by the discont inuati on of the use of the subtherapeutic doses of antib iot i cs in farm animals , or by discontinuation of use speci fically for growth promotion? The committee , in the discuss ion that fol low below , is incl ined to think that the total number of deaths due to salmone l losis would decl ine . However , these matters are not at present subj ect to scient i f ic proof . The committee did not deal with the convers ion o f drug­ susceptible bacterial organisms to drug-res istant clones ( by plasmid trans fer ) in s itu , but with the reduction in numbers of bacteria of drug-susceptible stra ins and the subsequent overgrowth with the more drug-resistant strains to f i l l the vacated ecologic niche . In thi s context , it may be useful to consider a s impl e diagram with two circles , one with deaths at some future t ime i f no discontinuation of antibiotics i s instituted , and one i f discontinuation has been put in pl ace as fol l ows : Deaths under D ea t h s i f aubt h e repe u t i c present po l i c i es , --+-- uses e r e banned bu t no t under a ben D ea t h s that wi l l occur e i t h e r way A ban then , would remove deaths in the left-hand lunule in thi s figure and replace them with by deaths in the r ight­ hand lunule . The committee might pose the quest ion about whether thi s shi ft is worth making . The committee has not attempted any risk-management pol icy analys i s ( it was not part of the committee ' s mandate ) , but bel ieves that the fol lowing comments are within its mandate . The l eft-hand lunule a lone is approximated , for various facets of the prob l em , by Figures VI I I - 1 through VI I I - 6 , and certa in aspects of the net d i f ference of the l eft-hand minus the right-hand lunule ( the best benefits of discontinuation o f antibiotics ) should b e approximated by F igures VI I I -7 through VI I I - 1 2 . The committee bel ieves that overal l there would be a net bene fit in reduced mortal ity ( thus , the right lunule might be sma l l er than the left ) , but th is cannot be proved with mathematical certainty , nor can the s i z e of the net bene fit be estimated with precis ion . Crit ical to cons iderat ion o f a ban is the l ikel ihood of a l ong-term reduction in the proportion of salmone l l a stra ins

172 with resi stance t o ant ibiotics . The genie is out of the bottle ; wil l it return? Res istant stra ins appear to have no survival advantage in the absence of cha l lenge by ant ib iotics ( otherwise they would have driven out susceptible strai ns long before the modern era ) , but there is l ittle evidence that they have a survival disadvantage either . Further , other uses of antibiotics wil l continue including therapy of infect ions in both humans and animals , and compl iance with a ban on subtherapeutic uses might be incomplete . Thus , it may be that a ban would retard the increase in proportion o f res istant stra ins , but not stop o r reverse the increase . BEFEBENCES 1. National Research Counc i l , Committee on the I nstitutional Means for Assessment of Risks to Publ ic Health . Risk Assessment in the Federal Government : Managing the Process . washington , D . C . : Nat ional Academy Press , 1 9 8 3 .

IX DIScuSSION The use of tetracycl ine and penic i l l in in subtherapeut ic concentrations in anima l and poultry feeds has aroused concerns about the poss ibi l ity of a risk to human health . There are good reasons for concern : the known properties o f trans ferable res i stance plasmids and transposons among bacteria , the powerful action of antimicrobial drugs in selecting for antimicrobial -resistant bacteria , and the high level s of antimicrobial res istance found in � � and salmone l l a i sol ates from farm animals and humans . In addit ion , it i s now poss ible t o detect clona l ly salmone l l a stra ins from various sources i n the food production cha in ( from farm to consumer) and so establ ish l inkages between isolates from humans and from farm anima l s ( or animal food products ) . Th is report dea l s ma inly with the magnitude of the human health hazard and with whether suf ficient data are ava i l able to assess the risk . There is no direct evidence to quanti fy the human health hazard from antibiot ic-res i stant pathogenic bacteria created by the use o f subtherapeutic amounts of pen ic i l l in or the tetracycl ines in animal feed . Us ing the ava i l able ind i rect evidence shows these ant ibiot ics in subtherapeutic concentrations do present a haz ard to human health and may contribute to a percentage ( see F igure VI I - 2 ) of the approximately 5 0 0 deaths annual ly in the United States from salmonel l os i s . Although the focus in the analys i s of risk has been only on deaths attributed to salmonel l os i s , there are the same concerns about risk due to � £Qli ( and other Enterobacteriaceae ) and to other pathogens ( both gram­ negative and gram-positive ) known to be drug res istant that might infect both anima l s and humans . Human exposure to enteric organisms ( pathogens and commensal s ) of an imal origin is extens ive . In food-animal process ing plants , the inc idence of bacterial contamination has been reported as high as 3 4 % for chickens , 7 4 % for beef , and 8 4 % for pork . 1 0 Figures reported for comparable � QQli contamination range from 7 3 % for beef carcasses , 8 1 % for chicken and 9 7 % for p ig carcasses . The � £Qli contamination presumably is from feca l sources . I n studies from Great Brita in , 7 , 8 3 8 % o f � QQli in cal f feces were res istant to one or more antimicrobial s , and other studies showed values of 4 9 % for pigs and 8 3 % for poultry . In the state o f Wash ington in surve i l l ance for enteric pathogens in a poultry process ing 17 3

Next: DISCUSSION »
Human Health Risks With the Subtherapeutic Use of Penicillin or Tetracyclines in Animal Feed Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF
  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!