AUTHOR(S): Patty Lovera , Felicia Nestor TITLE: Hamburger Hell: The Flip Side of S!"#s Salmo$ella Testi$g Program ABSTRACT This report e%ami$es the i$tegrity a$d reliability of the &S& !epartme$t of "griculture's (S!") Salmo$ella testi$g program& These microbial tests are the cor$ersto$e of its Ha*ard "$alysis Critical Co$trol Poi$t (H"CCP) program to moder$i*e food safety a$d tra$sform meat a$d poultry i$spectio$ i$to a public health program& +icrobial testi$g also serves as the policy ratio$ale for reduci$g gover$me$t visual i$spectio$ i$ slaughter a$d processi$g pla$ts& +ost visibly, it is the fou$datio$ for repeated age$cy stateme$ts that Salmo$ella co$tami$atio$ of gover$me$t,approved meat a$d poultry is droppi$g& -ased o$ a$ e%haustive, five mo$th revie. of S!"'s o.$ records, obtai$ed u$der the Freedom of /$formatio$ "ct, .e co$clude there is $o factual basis from the testi$g program for S!"'s reassura$ces that the food supply has become safer for co$sumers of grou$d beef& 0e e%ami$ed results from the start of the Salmo$ella testi$g program i$ 1a$uary 2334 through 5ctober 2, 6772& 0e fou$d that there has bee$ a systematic breakdo.$ i$ the i$tegrity of the sampli$g program, to the e%te$t that its results are u$able to support co$clusio$s about cha$ges i$ co$tami$atio$ rates& 0hether due to systematic i$compete$ce or bad faith, S!"'s 8do$'t look, do$'t fi$d8 policy mea$s it is fu$dame$tally deceivi$g the public .ith false reassura$ces& Please $ote that this article .as peer revie.ed but $ot double,bli$d revie.ed& ARTICLE "$ e%te$ded versio$ of this article is available at http:99...&citi*e$&org9docume$ts9salmo$ellareport&P!F
"'Who makes meat safe?' [she] asked. The crowd managed a weak, 'We do.' 'That's right,' Murano replied. 'You do. We [!"#] are $ust there to look o%er &our shoulder.' 'rom the (ack of the room, a %oice called out, ')an't &ou look awa& once in awhile?'"*+, 0hile 8do$'t look, do$'t fi$d8 regulatio$ may be a laughi$g matter for i$dustry i$siders, as this report .ill sho., it is defi$itely $ot a laughi$g matter for co$sumers& "t ma$y large grou$d beef pla$ts, the S!" is doi$g a poor :ob livi$g up to its promises about usi$g a $e., scie$tific, microbial testi$g program to protect co$sumers& The scale of food bor$e ill$ess i$ the &S&, estimated by the Ce$ters for !isease Co$trol a$d Preve$tio$ to cause ;< millio$ ill$esses, =6>,777 hospitali*atio$s a$d >,777 deaths per year(6), is staggeri$g& " stro$g microbial testi$g program could serve as a$ importa$t tool for reduci$g food bor$e ill$ess& -ut a poorly impleme$ted microbial testi$g system does $ot bolster federal food A Multinational Organisation's Approach to Multinational Coporations /$ rece$t times "m$esty /$ter$atio$al#s activities have e%te$ded to i$clude efforts to preve$t commercial orga$isatio$s from usi$g their fi$a$cial &&& Full Story ? Rigged Rules and Double Standards: Trade, Globalisation, and the Fight Against Poert! Trade is o$e of the most po.erful forces li$ki$g our lives, a$d a source of u$precede$ted .ealth& @et millio$s of the .orld#s poorest people are be &&& Full Story ? "thical Consu#eris# $ De#ocrac! through the %allet There are ma$y .ays co$sumers ca$ use their spe$di$g po.er to make a differe$ce to the .orld& This article discusses the co$cept of ethical co$sumeris &&& Full Story ? &'(C"F: Fro# Post$%ar "#ergenc! Aid to Child Rights )eader 5ver the past half ce$tury N/CAF has evolved from a$ emerge$cy body created to help childre$ i$ post,.ar Aurope to a leadi$g child rights orga$isa &&& Full Story ? Shri#p $ The Deastating Delicac! +a$y co$sumers are u$a.are of the impacts of shrimp farmi$g o$ both the people a$d the $atural e$viro$me$ts .here farmi$g occurs& This article prov &&& Full Story ? Consu#er Actiis# and Corporate Accountabilit! Blobal A%cha$ge is a huma$ rights orga$i*atio$ dedicated to promoti$g eco$omic, political, a$d social :ustice arou$d the .orld& Fou$ded i$ 2344, th &&& Full Story ? safety programs& "$d impleme$ti$g a scie$tific program i$ a less tha$ scie$tific ma$$er is a .aste of ta%payer dollars a$d a$ abuse of the public's trust&
*(STOR+ /$ 233=, -. coli,co$tami$ated hamburger killed four childre$ a$d sicke$ed more tha$ ;77 people $atio$.ide i$ the 1ack i$ the -o% outbreak(=)& The outbreak served as a .ake,up call a$d highlighted the severity of the problem of co$tami$ated food& /t also sparked a $atio$al debate o$ more effective a$d moder$ methods of gover$me$t food i$spectio$& Three years later, i$ 233<, the Cli$to$ "dmi$istratio$ fi$ali*ed the Pathoge$ Ceductio$9H"CCP (Ha*ard "$alysis Critical Co$trol Poi$t) rule, a modificatio$ of reDuireme$ts for federally i$spected meat a$d poultry slaughteri$g a$d processi$g establishme$ts& The primary cha$ge u$der H"CCP .as a shift i$ gover$me$t a$d i$dustry roles based upo$ the premise that i$troductio$ of microbial testi$g .ould stre$gthe$ meat i$spectio$ by maki$g it more scie$ce,based& $der H"CCP, pla$t ma$agers must a$aly*e their productio$ system for likely ha*ards, the$ ide$tify critical co$trol poi$ts .here these ha*ards are likely to occur& Pla$t employees must the$ co$trol a$d mo$itor the pla$t's process at these poi$ts& 0ith this $e. system, S!" tur$ed over ma$y i$spectio$ tasks traditio$ally performed by federal i$spectors to pla$t employees& Bover$me$t i$spectors, o$ce the backbo$e of product safety efforts, are $o. relegated, primarily, to oversight a$d auditi$g roles& Ceprese$tatives from some co$sumer orga$i*atio$s vie.ed the trade,off as acceptable because, for the first time, the S!" i$troduced routi$e gover$me$t microbial testi$g i$ meat pla$ts& 0he$ impleme$ted, this testi$g program co$sisted of a routi$e gover$me$t sampli$g program for !almonella*.,& The stated purpose of the !almonella testi$g program is $ot to preve$t products co$tami$ated .ith !almonella from leavi$g the pla$t& Cather, its purpose is to use the prese$ce of the pathoge$ as a$ i$dicatio$ that the pla$t is $ot co$trolli$g food safety ha*ards i$ their productio$ processes(>)& !almonella .as 8selected Eby the age$cyF as the target orga$ism because it is the most commo$ cause of food bor$e ill$ess associated .ith meat a$d poultry products8 a$d because the age$cy believed that co$trolli$g for !almonella might also result i$ reductio$s of other da$gerous pathoge$s that may gro. i$ the gut of farm a$imals, such as -. coli /+01231(<)& These pathoge$s ca$ get i$to the food supply .he$ sloppy ha$dli$g of the hides or i$testi$es of a$imals at slaughter allo.s fecal co$tami$atio$ of the meat& The age$cy claimed that !almonella sampli$g .ould more tha$ make up for dimi$ished fro$tli$e gover$me$t i$spectio$ a$d .ould make food safer by detecti$g da$gerous microbes that .ere i$visible to i$spectors& The age$cy .ould e$sure food safety by verifyi$g that pla$t productio$ systems .ere successfully co$trolli$g ha*ards& The e%plicit strategy .as that .he$ the microbial testi$g program detected food safety problems at a pla$t, the age$cy .ould ma$date that the pla$t take immediate corrective actio$s a$d .ould revie. the adeDuacy of those actio$s i$ a timely ma$$er(;)& S!" established !almonella performa$ce sta$dards at the begi$$i$g of H"CCP, a$d a$$ou$ced that these sta$dards .ould become progressively more rigorous, eve$tually leadi$g to a$ overall decrease i$ the pathoge$s o$ meat a$d poultry a$d, ultimately, a decrease i$ associated food,bor$e ill$esses& Compa$ies $ot able to meet the sta$dards .ould face regulatory actio$ ra$gi$g from .ritte$ $otices reDuiri$g corrective actio$ to .ithdra.al of i$spectio$& /f a pla$t failed three successive tests, the age$cy .ould suspe$d i$spectio$ services, effectively shutti$g the pla$t do.$, 8u$til the establishme$t demo$strates its ability to meet the performa$ce sta$dard&8(4) Ho.ever, i$ !ecember 6772, a Te%as federal appeals court ruled i$ Supreme -eef Processors, /$c& v& $ited States !epartme$t of "griculture(3), that the S!"'s !almonella performa$ce sta$dard is u$la.ful& The court's decisio$ applies o$ly to Te%as, +ississippi, a$d Louisia$a& "ccordi$g to S!"'s public affairs departme$t, the age$cy is usi$g the decisio$ to guide its policy $atio$.ide&(27) Still u$clear is e%actly ho. much of the !almonella testi$g program could be i$validated by the decisio$ that the sta$dard itself is u$la.ful& The age$cy has decided $ot to appeal the decisio$ to the Supreme Court(22), a$d has stated that .hile the decisio$ limited its ability to e$force the sta$dard, it did $ot dimi$ish its authority to use the performa$ce sta$dard as a diag$ostic tool(26)& $dersecretary for Food Safety !r& Alsa +ura$o has assured the public that 8Ethe (ntroduction Welcome to this special issue of the 4ournal of 5esearch for )onsumers. For this issue, $o$,profit orga$isatio$s .ere i$vited to co$ &&& Full Story ? *a#burger *ell: The Flip Side o, &SDA-s Sal#onella Testing Progra# This report e%ami$es the i$tegrity a$d reliability of the &S& !epartme$t of "griculture's (S!") Salmo$ella testi$g program& These microbial tests &&& Full Story ? Previous Issues Issue: 4, 677> Issue: ;, 677G Issue: <, 677= Issue: >, 677= Issue: G, 6776 Issue: 6, 6772 Issue: 2, 6772 age$cyF .ill co$ti$ue to test for !almonella, si$ce this .ill serve to verify that a Ecompa$y's food safetyF pla$ is properly co$trolli$g food,bor$e ha*ards to the greatest e%te$t possible&8 /$ fact, although there are curre$tly a $umber of proposals i$ Co$gress to give the age$cy i$creased authority to close pla$ts based o$ failed !almonella tests, !r& +ura$o rece$tly told co$gressio$al leaders that she .as $ot i$ favor of basi$g e$forceme$t decisio$s solely o$ microbial test results& /$stead, she believes they fu$ctio$ best to trigger more e%te$sive physical i$spectio$s(2=)& +ea$.hile, the age$cy is .aiti$g for pe$di$g recomme$datio$s from the Natio$al "cademy of Scie$ces a$d the Natio$al "dvisory Committee o$ the +icrobiological Criteria for Foods o$ ho. microbial sta$dards ca$ best be used for public protectio$(2G)& The age$cy is also pla$$i$g a series of forums to get public comme$t o$ this issue& 5bviously, i$ the .ake of the Supreme -eef decisio$, the future of the !almonella testi$g program is u$clear& 0hether or $ot S!" retai$s authority to take a$y e$forceme$t actio$ as a result of testi$g is a Duestio$ that must be resolved, most likely by Co$gress& For the purposes of this report, .e .ill $ot speculate o$ the ultimate effect of the decisio$& 0hat .e .ill do is e%ami$e the data ge$erated by the testi$g program from its i$ceptio$ through 5ctober, 6772 , a fe. mo$ths before the Supreme -eef decisio$ , a$d e%ami$e the program's effective$ess& This is useful because it ca$ i$form further discussio$s about the future of performa$ce sta$dard testi$g&
M"T*ODS 5$ 5ctober 2, 6772 .e submitted a Freedom of /$formatio$ "ct (F5/") reDuest to the S!" reDuesti$g all !almonella testi$g data collected si$ce the begi$$i$g of the testi$g program i$ 1a$uary 2334& The data .as received o$ C!,C5+ i$ !ecember 6772 a$d covered testi$g from 1a$uary 6<, 2334 through 5ctober 2, 6772& The grou$d beef data .as the$ separated from the data o$ other product categories& Brou$d beef testi$g data .as chose$ for a$alysis for the follo.i$g reaso$s: A$forci$g !almonella sta$dards i$ grou$d beef pla$ts .as the specific sub:ect of the rece$t Supreme -eef ruli$g& Brou$d beef testi$g is the o$ly routi$e scie$tific testi$g do$e at the largest beef pla$ts that both slaughter cattle a$d gri$d beef(2>)& -ecause .e .a$ted to evaluate age$cy e$forceme$t actio$s at pla$ts faili$g to meet the performa$ce sta$dard, .e sorted the data by the $umber of failed 8sample sets8 for each product type& The category .ith the most failed sets .as grou$d beef(2<), .hich makes it the largest body of data for evaluatio$ of the age$cy's e$forceme$t actio$s& !ata from the age$cy's ra$dom -. coli 52>;:H; testi$g program, a program i$depe$de$t from the !almonella testi$g program, i$dicates that the perce$tage of grou$d beef samples collected at federally i$spected productio$ facilities that tested positive for -. coli i$creased from &24H i$ 233; to 2&=H i$ 6777& Ave$ .ith better testi$g, a$ -. coli 52>;:H; rate e%ceedi$g o$e perce$t i$dicates that compa$y safety pla$s a$d curre$t age$cy regulatio$s are $ot co$trolli$g food safety ha*ards i$ a .ay that sufficie$tly protects co$sumers(2;)& Cece$t research i$dicates that co$sumer dema$d for beef is steadily i$creasi$g(24), a$d therefore the safety of this product greatly impacts public health& H"CCP .as impleme$ted i$ stages, accordi$g to pla$t si*e& "s pla$ts came u$der H"CCP, they simulta$eously became sub:ect to !almonella testi$g& H"CCP .as impleme$ted i$ large pla$ts (those .ith >77 or more employees) i$ 1a$uary 2334I i$ small pla$ts (those .ith 27 to G33 employees) i$ 1a$uary 2333I a$d very small pla$ts (those .ith fe.er tha$ 27 employees or a$$ual sales of less tha$ J6&> millio$) i$ 1a$uary 6777& /t is very difficult to determi$e ho. large a pla$t is strictly by the i$itial !almonella testi$g date, ho.ever, because testi$g at some large pla$ts started late (after 2334) .hile testi$g at some small pla$ts started early (before 1a$uary 2333)& The database .e received did $ot list pla$t si*e& Therefore .e first estimated a pla$t's si*e by the date it started !almonella testi$g& 0e co$firmed the si*e of the pla$ts .e believed to be large (.hich are the primary focus of this report) usi$g either the age$cy's .ebsite(23) or i$formatio$ .e received from age$cy officials& 0he$ this i$formatio$ co$flicted, .e relied o$ the .ebsite i$formatio$ because the age$cy .as freDue$tly $o$,respo$sive to our Duestio$s about pla$t si*e a$d other i$formatio$ perti$e$t to this report(67)& Co$fusio$ about pla$t si*e may also e$ter i$ because pla$t si*es are $ot fi%ed(62)& /t is possible that some pla$ts may have cha$ged category by gai$i$g or losi$g employees& The bulk of this report deals .ith fi$di$gs i$ the 6< grou$d beef pla$ts .e determi$ed to be large& -ecause the age$cy impleme$ted H"CCP i$ the large pla$ts first, it has do$e more testi$g i$ each of these pla$ts tha$ i$ pla$ts that are small or very small& There are, ho.ever, ma$y more small a$d very small pla$ts, so these co$tribute the ma:ority of tests i$ later years& Large pla$ts .ill have a sig$ifica$t impact o$ public health due to the large volume of grou$d beef they produce& There is $o reliable, publicly available i$formatio$ o$ the perce$tage of grou$d beef produced by large pla$ts(66), or o$ the total pou$dage produced by large grou$d beef pla$ts u$der H"CCP& For calculatio$s i$ this report, .e are usi$g a$ estimate commo$ly accepted by those familiar .ith the grou$d beef i$dustry that large grou$d beef pla$ts typically produce 2>7,777 pou$ds per shift, t.o shifts per day, si% days per .eek& -ased o$ this estimate, the 6< large grou$d beef pla$ts collectively se$d e$ough grou$d beef to market to make $early 27 billio$ Duarter pou$d hamburgers each year&
PART O'": T*" SALMONELLA T"ST('G PROGRAM O' PAP"R .S/ (' PRACT(C" T*" T"ST('G PROGRAM O' PAP"R The stated purpose of performa$ce sta$dard testi$g is to e$sure that food co$sumed by the public is as safe as possible& The age$cy uses a sampli$g program to verify that pla$t systems are co$trolli$g food safety ha*ards& The more freDue$t the sampli$g, the more rapidly the age$cy should be able to detect problems that threate$ public health& FreDue$t testi$g is $ecessary because the $ature of the productio$ systems i$ meat pla$ts makes them e%tremely vul$erable to problems that freDue$tly occur .ithout .ar$i$g , a$d as a result, food ca$ Duickly be tur$ed i$to a to%ic product& A%cessive li$e speeds preve$t li$e employees from detecti$g co$tami$atio$& Amployee tur$over i$ the large pla$ts ofte$ e%ceeds 277 perce$t per year& Thus, each $e. group of u$skilled .orkers must be trai$ed ho. to eviscerate a$imals .ithout spilli$g fecal material o$ the carcass, as .ell as ho. to avoid cross,co$tami$ati$g meat as it moves do.$ the li$e& ADuipme$t malfu$ctio$ or misha$dli$g ca$ also i$troduce a$d spread co$tami$atio$ o$ carcasses& Pla$ts that gri$d meats are vul$erable to problems .ith the ra. products they receive from their suppliers, as .ell as sa$itatio$ a$d process problems at their o.$ facility& 0he$ the public debate o$ microbial testi$g first bega$, co$sumer group represe$tatives advocated daily sampli$g at each pla$t& This seemed to be the o$ly .ay to determi$e that pla$ts .ere co$trolli$g ha*ards co$siste$tly e$ough to e$sure that the food .as actually safe& Some compa$ies that are kee$ly a.are of the deadly co$seDue$ces of i$adeDuate procedures use eve$ more freDue$t testi$g& "fter the 233= tragedy, 1ack,/$,The,-o% i$stituted microbial testi$g i$ its pla$ts every 2> mi$utes of productio$(6=)& The S!" regulates meat a$d poultry i$ slaughter a$d processi$g facilities& 8Slaughter8 pla$ts tur$ live a$imals i$to carcasses& "ll further butcheri$g, gri$di$g a$d processi$g of the carcasses is do$e at processi$g facilities& The age$cy's curre$t !almonella regulatio$s apply at slaughter pla$ts for cattle, hogs, chicke$ a$d turkeys, a$d processi$g pla$ts that gri$d beef, chicke$ or turkey, but $ot to other types of processed meats& 0hile the H"CCP Fi$al Cule authori*es the age$cy to sample carcasses a$d grou$d products at pla$ts that produce both(6G), the age$cy focuses testi$g efforts o$ grou$d products at these pla$ts because 8!almonella is more likely to be prese$t o$ ra., grou$d or commi$uted products tha$ o$ the carcasses from .hich they are derived&&&8(6>)& $der sta$dard protocol the age$cy tells federal i$spectors i$ a give$ pla$t .he$ to begi$ taki$g samples& For grou$d products, the sample si*e is appro%imately half a pou$d(6<)& "$ age$cy !irective i$structs the i$spectors to take the sample each day the pla$t is produci$g the product a$d to se$d the sample to o$e of the S!" labs for a$alysis(6;)& For each product that S!" tests, it has established the $umber of samples that comprise a complete sample 8set&8 0he$ the lab has a$aly*ed the desig$ated $umber of samples, it i$forms the i$spector i$ the pla$t to stop se$di$g them& The results of the lab tests are se$t to the Food Safety /$spectio$ Service (FS/S) at S!" headDuarters i$ 0ashi$gto$, !C& 5$ly .he$ the sample set is complete, do the gover$me$t i$spector i$ the pla$t or pla$t officials have access to the results& !almonella sta$dards for each product .ere based o$ $atio$.ide i$dustry baseli$e testi$g performed before the impleme$tatio$ of H"CCP& -aseli$e prevale$ce rates .ere calculated usi$g data from large a$d small pla$ts o$ly, because very small pla$ts co$tribute less tha$ o$e perce$t of products reachi$g the market(64)& For each product, the age$cy established the $umber of samples that make up a set a$d the $umber of samples i$ the set that may legally test positive for the prese$ce of !almonella& The performa$ce sta$dards established by the age$cy .ere chose$ so that there is a$ 8appro%imately 47H probability of passi$g .he$ the establishme$t is operati$g&&&:ust .ithi$ the performa$ce sta$dard&8(63) The $umber of allo.able positives a$d the $umber of samples i$ the set vary by product& For e%ample, a pla$t slaughteri$g steers a$d heifers meets the sta$dard if $o more tha$ o$e out of 46 samples tests positive for !almonella, .hereas a grou$d turkey pla$t meets the sta$dard if $o more tha$ 63 out of >= samples tests positive for !almonella& Brou$d beef pla$ts, covered i$ this report, are co$sidered to be i$ complia$ce if $o more tha$ five out of a set of >= samples test positive for !almonella(=7)& /f the pla$t passes the set, it .ill be tested agai$ .he$ever the age$cy schedules it& Ho.ever, the age$cy has $ever made clear ho. soo$ after passi$g a set the pla$t .ill be re,tested, thereby $eglecti$g to establish a mi$imum scope or freDue$cy of testi$g& This leaves the public health fou$datio$ of the testi$g program i$complete& /f the pla$t fails the set, a$ i$spector i$ the pla$t is i$structed to issue a $o$, complia$ce report reDuiri$g corrective actio$& The$, the age$cy .ill 8$ormally8 begi$ re,testi$g i$ appro%imately <7 days to verify that the cha$ges impleme$ted by the pla$t have bee$ effective(=2)& -efore the rece$t Supreme -eef decisio$ prohibited it, age$cy regulatio$s provided that further failures could lead to a shut do.$ after three successive failed sets,(=6) but $o. these serious lapses i$ food safety may o$ly trigger a$ 8e%te$sive revie.8 by the age$cy& (See page 2< for more discussio$ of the revie. process&) /t is importa$t to $ote that by the time a set is complete, the product that .as tested .ill have left the pla$t, a$d the !almonella testi$g program has $o ma$date or e%pectatio$ that product from pla$ts faili$g to meet the sta$dard .ill be recalled& T*" T"ST('G PROGRAM (' PRACT(C" There are appro%imately 82,;77 establishme$ts produci$g grou$d beef u$der ES!"F i$spectio$8(==)& The database received a$d a$aly*ed for this report has 2,2<= complete !almonella testi$g sample sets from ;<7 federally i$spected pla$ts that produce grou$d beef& T.e$ty,si% of these pla$ts are large, ;=G are small or very small& 5f the 6< large pla$ts, 2< $ever failed a set (<2&> perce$t)& 5f the 27 that failed a$y sets: K 5$e pla$t failed its o$ly set (292)& K 5$e pla$t failed o$e out of t.o sets (296)& K Three pla$ts failed o$e out of three sets (29=)& K 5$e pla$t failed t.o out of three sets (69=)& K 5$e pla$t failed three out of three sets (=9=)& K T.o pla$ts failed o$e out of four sets (29G)& K 5$e pla$t failed t.o out of four sets (69G)& 5f the ;=G small a$d very small pla$ts, <<< $ever failed a set (37&; perce$t)& 5f the <7 small pla$ts that failed a$y sets: K Seve$ pla$ts failed their o$ly set (292)& K 2> pla$ts failed o$e out of t.o sets (296)& K 6= pla$ts failed o$e out of three sets (29=)& K Four pla$ts failed t.o out of t.o sets (696)& K Five pla$ts failed t.o out of three sets (69=)& K Three pla$ts failed t.o out of four sets (69G)& K 5$e pla$t failed three out of four sets (=9G)& K 5$e pla$t failed three out of five sets (=9>)& K 5$e pla$t failed four out of four sets (G9G)& 5f the eight very small pla$ts that failed a$y sets: K T.o pla$ts failed their o$ly set (292)& K Five pla$ts failed o$e out of t.o sets (296)& K 5$e pla$t failed t.o out of t.o sets (696)& Testing Fre0uenc! "ge$cy protocol states that a sample is to be collected every day a pla$t produces grou$d beef u$til >= samples are collected(=G)& To determi$e the e%te$t to .hich the age$cy actually samples grou$d beef each day that it is bei$g produced, .e looked o$ly at data from the large pla$ts, because these pla$ts typically produce grou$d beef five or si% days per .eek& Small a$d very small pla$ts may o$ly produce the product sporadically& /f samples are take$ daily, the age$cy should easily complete collectio$ of a grou$d beef sample set at a large pla$t .ithi$ 3 to 26 .eeks& "ge$cy testi$g .as completed .ithi$ a 26,.eek time frame i$ less tha$ G7H (=3&;H) of the ;= sets completed i$ the large pla$ts for .hich .e have good data(=>)& The media$ set le$gth .as 2G&2 .eeks a$d the ra$ge of set le$gths .as 27&= .eeks to 26<&G .eeks i$ large pla$ts& -ecause of the e%treme le$gth of a set at a $umber of pla$ts, the average le$gth of a sample set at large pla$ts .as 6>&; .eeks& The data sho. that prolo$ged sets ofte$ have i$volved e%te$sive lapses .ithi$ the sample set, rather tha$ co$siste$t, but less freDue$t testi$g& "$y $umber of the typical tra$sie$t problems i$volvi$g eDuipme$t, supplies or employees could combi$e to cripple food safety systems duri$g such a lapse, a$d remai$ u$detected if the pla$t co$ti$ues $ormal productio$ schedules a$d co$ti$ues to se$d product to the market(=<)& 6n the eight large plants where a set took more than a &ear, all sampling ceased for periods ranging from nine to nearl& 78 months, completel& precluding the go%ernment from detecting pro(lems during these lapses. These lo$g lapses i$ testi$g raise a$other co$cer$& /$ respo$se to our Duestio$s, the age$cy has repeatedly stated that delays i$ completi$g sample sets are most ofte$ due to i$dividual circumsta$ces such as sample delivery problems or other u$foresee$ logistical delays& The data, ho.ever, sho. stro$g patter$s of delay that call such e%pla$atio$s i$to Duestio$& /$ the eight pla$ts me$tio$ed above, the age$cy started the sets i$ 2334 a$d sampled $early daily u$til most of the >= samples .ere collected& The$ all sampli$g ceased u$til the fi$al fe. samples $ecessary to complete the set .ere collected i$ 2333 or 6777& -ased o$ the age$cy's respo$ses to our Duestio$s, it is $ot clear if the age$cy k$e. of the lapses i$ testi$g, or a$y pote$tial patter$ i$ the timi$g of these lapses& Cegardless of the reaso$ behi$d it, the age$cy's failure to preve$t these lapses i$ testi$g is i$e%cusable& Postponing a Finding o, Failure Si$ce a grou$d beef pla$t fails the sample set if more tha$ five samples test positive for !almonella, .e refer to the date of the si%th positive sample as the 8actual failure8 date& Ho.ever, as this sectio$ e%plai$s, the age$cy does $ot declare a set a failure u$til all samples are collected& 0e refer to the date that the age$cy co$siders a set a failure as the 8official failure8 date& The age$cy ma$dates that pla$ts take corrective actio$ after faili$g a set(=;), but it u$$ecessarily postpo$es determi$atio$ of failure a$d thereby prolo$gs co$ditio$s that could threate$ public health& Ave$ though a grou$d beef pla$t fails if a$y more tha$ five samples test positive for !almonella, the age$cy does $ot co$sider a set complete a$d .ill $ot address set failure u$til all >= samples have bee$ collected& The age$cy does have a$ 8early .ar$i$g8 system, u$der .hich it i$forms pla$ts as soo$ as they have accumulated more tha$ half the positives $ecessary to fail a sample set (three for grou$d beef)& -ut the age$cy does $ot reDuire that the pla$t take a$y actio$ to correct the problem after such a .ar$i$g& "fter this early .ar$i$g, u$til all >= samples are a$aly*ed, the age$cy .ill $ot: K i$form the pla$t .he$ it fails the set (.ith the si%th positive sample)I K reDuire corrective actio$I K alert the i$spectors i$ the pla$t that the pla$t failed the setI or K .ar$ the public that the pla$t may be se$di$g u$usually co$tami$ated products to market& /$stead, the S!" does $othi$g a$d co$ti$ues to stamp the products 8/$spected a$d Passed8 a$d the product goes to market& The delays due to prolo$ged testi$g discussed i$ the previous sectio$ are especially sig$ifica$t at faili$g pla$ts& 5ut of the 2G faili$g sample sets at large pla$ts, o$ly four .ere completed i$ 26 .eeks& /$ the other 27 faili$g sets, testi$g took up to 6; additio$al .eeks& -ecause of the e%cessive time to complete the set, a$d because $o corrective actio$ is reDuired u$til a set is complete, such delays represe$t a missed opportu$ity to protect public health& The ultimate impact of this dela& on consumers is that the agenc& allowed failing plants to send products to market for a cumulati%e total of nearl& +999 weeks after actual failures, without informing them of pro(lems or re:uiring them to take correcti%e action. "uring the +7+ weeks of dela&s at large plants alone, the agenc& knowingl& allowed an estimated 7+; million pounds of potentiall& contaminated ground (eef to enter the market (earing the !"# seal of appro%al (efore it e%en informed plant managers of the need for correcti%e actions*<;,. *=round (eef was also sent to the market during the waiting period that followed these sets., This u$$ecessary delay .he$ corrective actio$s are $eeded most poses a treme$dous threat to co$sumers by postpo$i$g elimi$atio$ of public health ha*ards, especially at pla$ts .here high $umbers of positive samples i$dicate that food safety systems are seriously compromised& For e%ample, at Co$"gra -eef Compa$y (Cactus, Te%as), over 2< .eeks passed after its si%th positive sample .as discovered by the age$cy a$d before corrective actio$ .as reDuired& "t A%cel Corporatio$ (-oo$eville, "rka$sas), over 23 .eeks passed after its si%th positive sample .as discovered by the age$cy a$d before corrective actio$ .as reDuired& -y the time all >= samples .ere collected i$ each set, the Co$"gra pla$t had 6> samples a$d the A%cel pla$t had 2> samples that had tested positive for !almonella, demo$strati$g that their food safety co$trols .ere seriously compromised throughout the u$$ecessary period of age$cy i$actio$& The %aiting Period 1et2een Sets 0e calculated the le$gth of the period bet.ee$ the e$d of o$e sample set a$d the begi$$i$g of the $e%t set& 0e refer to this period as a 8.aiti$g period&8 The age$cy does $ot defi$e ho. lo$g the .aiti$g period is after passi$g sets& Though $o testi$g is bei$g do$e duri$g a .aiti$g period, grou$d beef co$ti$ues to carry the S!" seal of approval& There .ere limitatio$s i$ the data that limited our ability to calculate the average .aiti$g period after passi$g sets& 0e reDuested data from 1a$uary 2334 through 5ctober 6772, but, u$der age$cy policy, o$ly received data from sample sets that had bee$ completed i$ that timeframe& There .ere $umerous pla$ts .here at least o$e set .as take$ but .here $o subseDue$t set has bee$ docume$ted si$ce 2333& 0hile it is possible that $o testi$g of grou$d beef has occurred at a particular pla$t si$ce the$, it is also possible that the age$cy had i$itiated but $ot completed a set by 5ctober 6772 a$d therefore did $ot release data from these i$complete sets& /t is also possible that the pla$t has stopped produci$g grou$d beef altogether& There .ere too ma$y cases like this to be i$dividually i$vestigated& Bive$ those limitatio$s, ho.ever, .e determi$ed the .aiti$g period after passi$g sets for pla$ts for .hich .e have records& The .aiti$g period at pla$ts that $ever failed a set e%hibited variability at both large a$d smaller pla$ts& # se%en or eight month waiting period for plants that ha%e ne%er failed a set can still pose a threat to pu(lic health (ecause good performance does not alwa&s continue. This is illustrated (& the fact that si> of the +9 large plants and +< of the 89 small plants that failed their second or third set, did so after passing their first set. Ave$ at pla$ts that pass co$secutive sets, there is $o assura$ce that the co$ditio$s duri$g the .aiti$g period bet.ee$ sets .ere co$sta$t& 5$e e%ample is the A%cel pla$t i$ Fort +orga$, Colorado& Numerous food safety problems .ere evide$t at this pla$t duri$g the test,free .aiti$g period bet.ee$ "pril 2>, 2333 a$d 1uly 24, 6777(=3): K 5$ September 2=, 2333, S!" fou$d -. coli 9+01321 i$ beef produced at the pla$t through its ra$dom retail sampli$g programI K S!" tests fou$d the same pathoge$ t.ice soo$ afterI K i$spectors cited the pla$t 6< times for fecal co$tami$atio$I K S!" suspe$ded i$spectio$ at the pla$t for three days a$d threate$ed a seco$d suspe$sio$I K sirloi$ tips produced at the pla$t o$ 1u$e 6=, 6777 a$d co$tami$ated .ith the deadly -. coli /+01321 pathoge$ .ere li$ked to a$ outbreak that sicke$ed over >77 people a$d killed a little girl i$ 0isco$si$&(G7) Testi$g of the grou$d beef .as the o$ly !almonella testi$g do$e at this e$tire pla$t , i$ either slaughter or processi$g& !espite the severity of the problems :ust described that .ere happe$i$g i$ the slaughter sectio$ of the pla$t starti$g i$ "pril 2333, the age$cy did $ot schedule a$other set to evaluate the pla$t's food safety co$trols u$til 1uly 6777& Lirtually all of the beef used by the pla$t to produce grou$d beef comes from the slaughter sectio$ of this facility& 0he$ a pla$t fails a sample set, age$cy policy holds that a follo.,up set .ill usually be i$itiated .ithi$ <7 days (appro%imately 3 .eeks)& "lthough the age$cy does $o testi$g duri$g this period, it co$ti$ues to place the S!" seal of approval o$ products leavi$g the pla$t& No$e of the pla$t si*e categories .ere fou$d to have media$ .aiti$g periods less tha$ the recomme$ded <7 days, moreover the data sho. that large pla$ts have bee$ give$ more time tha$ small or very small pla$ts after a failed set before the age$cy resumes testi$g for !almonella& This creates a$d prolo$gs a$ i$terim ho$or system .he$ S!" should i$stead be i$te$sifyi$g gover$me$t oversight& "gai$, because of the volume of grou$d beef se$t to market by large pla$ts, it is of particular co$cer$ for co$sumers .he$ the age$cy delays verificatio$ of pla$t corrective measures at large pla$ts& $der the !almonella testi$g program, the o$ly .ay to get such verificatio$ is to complete a passi$g sample set& The Final Dela! Cece$tly, S!"'s $dersecretary for Food Safety !r& Alsa +ura$o stated that, i$ respo$se to a lo.er court's decisio$ i$ the Supreme -eef case i$ 2333, the age$cy cha$ged its protocol for scheduli$g a third sample set(G2)& This cha$ge .as formali*ed i$ 5ctober 6772, .he$ the age$cy issued a !irective stati$g that after a pla$t fails t.o co$secutive !almonella performa$ce sta$dard sets, the age$cy .ill $o lo$ger begi$ a follo.,up set after <7 days& /$stead, it .ill se$d i$ a group of age$cy officials to assess 8.hether a$ establishme$t is carryi$g out activities that meet the reDuireme$ts of the Pathoge$ Ceductio$9H"CCP regulatio$&8(G6) The evaluatio$ is called a$ 8/$,!epth Lerificatio$ (/!L) revie.&8 Products co$ti$ue to carry the S!" seal of approval to market throughout the revie.& -ut u$fortu$ately for co$sumers, age$cy policy provides a$ e%cuse for $ot starti$g the third sample set& "ccordi$g to age$cy policy, if there is 8a$y doubt8 that pla$t corrective actio$s after a failure have bee$ effective or a$y doubt that the pla$t 8is likely to pass the third set,8 the third !almonella set .ill $ot begi$& /t is o$ly .he$ there is a 8high level of co$fide$ce that&&&the $e%t sample set .ill succeed&&&EthatF the third sample set is i$itiated&8(G=) /$ short, u$less the age$cy is sure the pla$t .ill pass, it .ill $ot begi$ a follo.,up set& The /!L revie. team may eve$ use the results of !almonella testi$g by the pla$t to determi$e .he$ the pla$t is likely to pass the set, esse$tially gra$ti$g the pla$t the cha$ce for off,the,record testi$g& Cepeatedly, i$ a$s.er to our Duestio$s, officials at the Food Safety /$spectio$ Service's Tech$ical Service Ce$ter i$ "mes, /o.a a$d at 0ashi$gto$ headDuarters said there is $o time limit to i$itiate a$other sample set duri$g a$ /!L& -oth li$e i$spectors a$d FS/S supervisors .ho are $o. .histleblo.ers have told us that this process has already go$e o$ for almost a year at some pla$ts& /t is already clear that /!L revie.s .ill o$ly :ustify more delays i$ taki$g actio$ to preve$t pote$tially da$gerous products from e$teri$g the marketplace& The ",,ect o, Multiple Dela!s Perhaps the .eak$esses of the !almonella testi$g program are best illustrated by co$sideri$g ho. little testi$g the age$cy does i$ a pla$t .he$ the various delays :ust described are combi$ed& 0e calculated the time bet.ee$ the first sample of a set a$d the first sample of the follo.i$g set a$d are calli$g this period the 8testi$g cycle&8 The testi$g cycle e$compasses the time to complete a full sample set a$d the .aiti$g period after that set& The testi$g cycle calculatio$ illustrates the cumulative effect of prolo$ged sample sets a$d .aiti$g periods discussed previously& The media$ testi$g cycle for sample sets at large grou$d beef pla$ts that have $ever failed a set is >4&> .eeks(GG), a$d the ra$ge .as 62&< .eeks to 2=<&; .eeks& /$cluded i$ these are five sets for .hich the testi$g cycles took more tha$ t.o years to complete& (/t is importa$t to keep i$ mi$d that i$ o$e testi$g cycle for grou$d beef, o$ly >= samples are take$&) The cumulative effect of such delays is less tha$ .eekly sampli$g, a sig$ifica$t departure from early FS/S pla$s for daily testi$g&(G>) 0hile it is impossible to li$k lo$g testi$g cycles directly .ith public health effects, a lo$g testi$g cycle $ecessarily co$tai$s lo$g periods .ith $o sampli$g& /t is $ot safe to assume that good performa$ce co$ti$ues duri$g test,free periods, .hether occurri$g duri$g passi$g sets or i$ the follo.i$g .aiti$g period& The data sho. t.o pla$ts .ith lo$g testi$g cycles .hich passed their first set, but .hose subseDue$t history illustrates .hy such delays are troubli$g& !ue to a sig$ifica$t lapse .ithi$ its first set, as .ell as the .aiti$g period after that set(G<), the /-P pla$t i$ Holcomb, Ma$sas o$ly had eight samples take$ bet.ee$ "ugust 2334 a$d 5ctober 6777& The pla$t .e$t o$ to effectively fail its seco$d set after o$ly 26 samples .ere a$aly*ed a$d fi$ished that set .ith almost t.ice the allo.able $umber of positives , 3 (G;)& !ue to a sig$ifica$t lapse .ithi$ its first set, as .ell as the .aiti$g period after the set(G4), the /-P pla$t i$ Amporia, Ma$sas had o$ly si% samples take$ bet.ee$ September 2334 a$d 1uly 6777& The pla$t .e$t o$ to effectively fail its seco$d set after o$ly 24 samples .ere a$aly*ed a$d fi$ished the set .ith more tha$ t.ice the allo.able $umber of positives (22)& For both of these pla$ts, the results i$dicate that the food safety systems .ere severely compromised by the time the age$cy fi$ally bega$ the seco$d setI ho.ever, it is $o. impossible to determi$e .he$ duri$g the test,free period either pla$t started posi$g a$ i$ordi$ate risk to public health& "t faili$g pla$ts, delays ca$ combi$e to create serious public health co$seDue$ces& " revie. of the sampli$g summaries from t.o large pla$ts that failed multiple sets demo$strates ho. prolo$ged sets, lo$g lapses bet.ee$ actual a$d official failure, a$d lo$g .aiti$g periods bet.ee$ sets combi$e to preve$t the !almonella testi$g program from bei$g used to effectively protect the public& The age$cy first determi$ed that pla$t systems at a Co$"gra pla$t .ere i$adeDuate o$ 1u$e, 27 2333& /t the$ took over 2< .eeks to complete the set i$ .hich the outcome .as already k$o.$, a$d over =G .eeks to start a seco$d set, .hich the pla$t proceeded to fail .ithi$ four .eeks& -ut as of 5ctober 2, 6772, it had $ot yet completed a passi$g sample set that could verify that problems had bee$ corrected& Though this plant had the worst record of an& large ground (eef plant in the data(ase, !"# has continued to stamp their product "6nspected and ?assed," with no ma$or interruptions as far as can (e determined.*.@, /$ the case of a$ A%cel pla$t, the age$cy first determi$ed that pla$t systems .ere i$adeDuate o$ +ay 6;, 2333& -ut as of 5ctober 2, 6772, it had $ot yet completed a passi$g sample set that could verify that corrective actio$s .ere effective& -%er& week that the agenc& allowed *or continues to allow, these two plants to send ground (eef to market without %erif&ing that the& ha%e successfull& corrected food safet& pro(lems, it ga%e the !"# seal of appro%al to enough potentiall& contaminated ground (eef to make o%er +. million :uarterApound ham(urgers.*09, 6f these two plants operated %irtuall& without interruption (etween their failures in 4une +@@@ and /cto(er +, 799+, together, the& introduced enough potentiall& dangerous ground (eef into the market to make nearl& 7 (illion :uarterApound ham(urgers, all !"#Aappro%ed.
PART T%O: A'A)+S(S OF &SDA's SALMONELLA T"ST('G R"PORTS /$ Part 5$e, !almonella testi$g data .as a$aly*ed to e%ami$e age$cy impleme$tatio$ of the testi$g program& /$ this sectio$ the data from large grou$d beef pla$ts .as a$aly*ed to determi$e .hat it i$dicated about improveme$ts i$ food safety co$trols i$ these pla$ts a$d about the accuracy of age$cy reports o$ the program& -ut u$fortu$ately, the data .ere so compromised by biases resulti$g from the poor impleme$tatio$ a$d desig$ of the program that reliable co$clusio$s about improveme$ts o$ the basis of the !almonella data ca$$ot be dra.$& The S!", ho.ever, has used this same data set to claim ma:or improveme$ts i$ food safety due to impleme$tatio$ of the H"CCP program& T*" 3&A)(T+ OF AG"'C+ DATA Bive$ the critical role of the testi$g program i$ evaluati$g H"CCP's effective$ess, it is reaso$able to e%pect that the program .ould be carefully pla$$ed a$d admi$istered& From the begi$$i$g of H"CCP impleme$tatio$, the results of the !almonella testi$g program .ere vie.ed as the primary a$d most reliable i$dicator for the effective$ess of the age$cy's meat a$d poultry i$spectio$ system& The age$cy stated that: "!almonella enforcement strateg& will em(od& an o($ecti%e, uniform s&stems approach to ensure that it is administered and applied in a fair, e:uita(le, and commonAsense manner. The #genc& will carefull& monitor and ad$ust its enforcement program on an ongoing (asis to ensure that its enforcement acti%ities reflect these principles while ensuring food safet&."*0+, Part 5$e demo$strated that testi$g .as $ot admi$istered i$ a u$iform .ay, resulti$g i$ lo$g delays both .ithi$ a$d bet.ee$ sets, a$d docume$ted e$ough i$co$siste$cies to cast doubt o$ the ra$dom$ess of testi$g results& Ca$dom$ess is a$ esse$tial compo$e$t of a$y legitimate sampli$g program& The !almonella program's e%treme i$co$siste$cies could have sig$ifica$tly biased data collectio$ u$der this program, precludi$g reliable a$alysis of a$$ual tre$ds or tre$ds bet.ee$ sample sets& T*" 3&A)(T+ OF AG"'C+ R"PORTS The age$cy has a$aly*ed o$,goi$g data from the program o$ at least eight separate occasio$s(>6)& Aach time it published at least o$e report o$ the fi$di$gs .ith accompa$yi$g press releases& /$itial age$cy reports stated that results .ere still co$sidered prelimi$ary a$d some reports suggest cautio$ i$ compari$g data because of pote$tial bias& !espite this, the age$cy co$siste$tly makes compariso$s to pre,H"CCP baseli$e data a$d .ith every a$alysis reports co$ti$ui$g improveme$t i$ the i$dustry as a .hole& The age$cy has ofte$ cited particular improveme$ts i$ grou$d beef as a result of H"CCP impleme$tatio$, based o$ the results from the !almonella testi$g program& The validity of these reports is severely compromised by t.o factors& First, the age$cy did $ot co$trol for pote$tial biases i$ the data due to poor impleme$tatio$ of the program& Seco$d, the age$cy used a$alytic methods that i$troduced additio$al bias& "ge$cy reports do $ot me$tio$ the great disparities i$ sample set le$gth a$d sampli$g patter$s or the pote$tial departure from a ra$dom sampli$g scheme that these disparities sig$al& There is $o i$dicatio$ that the age$cy recog$i*ed this source of bias or co$trolled for it& The age$cy's a$alytic methods for i$terpreti$g the data add additio$al bias& The t.o measures of performa$ce co$siste$tly used by the age$cy i$ its reports are complia$ce a$d prevale$ce rates& The 8complia$ce rate8 i$dicates the perce$tage of completed sample sets that passed the performa$ce sta$dard duri$g a give$ time period& The 8prevale$ce rate8 i$dicates the perce$tage of samples that tested positive for !almonella& /$ all four reports si$ce 1a$uary 6777, the age$cy has e%cluded results from some sample sets from most complia$ce rate a$d prevale$ce rate calculatio$s& These calculatio$s i$clude results o$ly from completed 8"8 sets, defi$ed as a$y set .hich does $ot follo. a failed set&(>=) (/t should be $oted that e%clusio$ of data from sets follo.i$g failed sets i$ $o .ay i$dicates that a$y of the meat from the sets is preve$ted from carryi$g the S!" seal to market&) 0he$ the age$cy i$itially used this method i$ calculatio$s for the report o$ 1a$uary 2334 through 1a$uary 6777 data (herei$after, 8the 2334,6777 report8), it did $ot me$tio$ it i$ the mai$ report(>G) or its accompa$yi$g press release&(>>) Complia$ce rates a$d prevale$ce rates .ere prese$ted as if $o data had bee$ e%cluded& The age$cy did describe the e%clusio$s i$ the te%t of the full report but prese$ted the results of calculatio$s .ith all the data o$ly i$ the "ppe$di% of that report(><)& The i$clusive (.hich did $ot use the " set method) results are dramatically differe$t& For grou$d beef, the prevale$ce rates do $ot i$dicate improveme$t, .hile the e%clusive results .hich are give$ sho. improveme$t& The baseli$e prevale$ce rate for grou$d beef is ;&>H a$d this figure is used to demo$strate improveme$ts due to H"CCP& /$ the 2334,6777 report, the large grou$d beef prevale$ce rate for sample sets completed i$ the seco$d year usi$g " sets o$ly, .as >&4 perce$t& This .as the figure used i$ all public a$$ou$ceme$ts about the testi$g program& /$ co$trast, the prevale$ce rate for all sets completed i$ the seco$d year i$ these pla$ts, .hich .as prese$ted o$ly i$ the "ppe$di% of the report, .as ;&< perce$t& 5$ly by usi$g the " set calculatio$ method could the age$cy report improveme$ts i$ the prevale$ce rate at large grou$d beef pla$ts i$ the seco$d year of H"CCP& The age$cy says that it i$cludes results o$ly from " sets because all others represe$t biased results a$d that " set data 8provides the most direct compariso$ to the baseli$e estimates used to establish the performa$ce sta$dards8&(>;) Ho.ever, e%cludi$g all but " sets from age$cy calculatio$s is certai$ to bias the data because a faili$g grou$d beef pla$t .as more tha$ = times (=&66) (>4) as likely to fail the $e%t set as a passi$g grou$d beef pla$t, i$ the data .e received& The age$cy has progressively 8purified8 the data by e%cludi$g results from these pla$ts for its reports& For e%ample, i$ 6777, five of the 6; sets collected i$ large grou$d beef pla$ts failed the performa$ce sta$dard& This mea$s that results from the five subseDue$t sets at those pla$ts .ould be e%cluded from subseDue$t calculatio$s& "ssumi$g that the age$cy used a similar si*e sample frame for the $e%t set, $early 67H of results (>96; sets) .ould be e%cluded from calculatio$s i$volvi$g the subseDue$t sets& There are t.o more areas of co$cer$ .ith the age$cy reports , the sampli$g program's relatio$ship to volume of productio$ a$d the diluti$g effect of i$cludi$g large $umbers of results from small pla$ts& $less the perce$tage of samples from each si*e category used i$ age$cy calculatio$s accurately represe$ts the volume of grou$d beef those pla$ts co$tribute to the market, the combi$ed prevale$ce rate .ill $ot reflect the prevale$ce of !almonella i$ grou$d beef o$ the market& /$ its reports, the age$cy has $ot co$trolled for this pote$tial bias a$d yet claims that the 8EdFata released& & &sho.s that the prevale$ce of !almonella i$ ra. meat a$d poultry has decreased&&&8(>3)& /$ grou$d beef calculatio$s, the bias due to this i$accuracy has bee$ co$siderable& For e%ample, i$ the 2334,6772 report $early =7 perce$t (<7) of samples used for grou$d beef prevale$ce calculatio$s for the 2334,6772 aggregate figure came from very small pla$ts although they produce less tha$ o$e perce$t of product o$ the market&(<2) +ore importa$tly, because very small pla$ts did so much better tha$ the large pla$ts i$ terms of !almonella reductio$s(<6), the fact that they co$tributed five times(<=) the $umber of samples as large pla$ts to the aggregate figure mea$s that the fi$al prevale$ce rate prese$ted by the age$cy (=&GH) is i$ large part due to 8dilutio$8 of the poor large pla$t data by the more successful small a$d very small pla$ts as they .ere added to the program over time&(<G) Had the age$cy e%cluded the samples from very small pla$ts from the aggregate figure a$d used the figures for large a$d small pla$ts o$ly, the prevale$ce rate .ould have bee$ =&3H& This result .ould be similarly i$accurate, ho.ever, u$less the perce$tage of small a$d large pla$t samples accurately reflected the volume of product they co$tribute to the market& -ecause of the biases i$troduced by the age$cy's a$alytical methods, the !almonella results touted by age$cy reports are very u$likely to represe$t the true co$ditio$ of grou$d beef o$ the market, yet it is the o$ly i$formatio$ available to the public a$d is eve$ used i$ Co$gressio$al testimo$y& /$ September 6777, the$, Secretary of "griculture Blickma$ used biased figures i$ his Stateme$t -efore the Se$ate Committee o$ "griculture, Nutritio$ a$d Forestry&(<>) He testified that 8E.Fith the Pathoge$ Ceductio$ a$d H"CCP rule, the prevale$ce of !almonella o$ ra. products has bee$ substa$tially reduced,8 a$d he reported a >&7H prevale$ce o$ grou$d beef& This figure .as the result of usi$g o$ly 8"8 sets a$d i$cludi$g $early 2> times as ma$y samples from small grou$d beef pla$ts as from large pla$ts&(<<) Neither biasi$g factor .as me$tio$ed to Co$gress&
PART T*R"": CO'C)&S(O'S 5ur a$alysis of the S!"'s !almonella performa$ce sta$dard program led to the follo.i$g co$clusio$s: 2& 0hile sampli$g is supposed to happe$ o$ a daily basis, it is ofte$ happe$i$g much less freDue$tly, .ith lo$g test,free periods prolo$gi$g sample sets& 6& The age$cy misses opportu$ities to protect co$sumers due to its policy of completi$g a sample set before reDuiri$g corrective actio$, eve$ if a pla$t has already failed a$d eve$ though delays i$ set completio$ are commo$& =& Lo$g .aiti$g periods bet.ee$ sample sets create additio$al a$d substa$tial test, free periods& G& The program .as origi$ally desig$ed as a 8three strikes a$d you're out8 performa$ce sta$dard& -ut the age$cy practice of i$defi$itely postpo$i$g a sample set after t.o failures creates a$d prolo$gs a$ i$dustry ho$or system at times. .he$ gover$me$t oversight should i$te$sify& >& 0he$ combi$ed, all the delays i$ the testi$g program create lo$g periods .he$ $o !almonella testi$g is performed& @et meat is stamped 8/$spected a$d Passed8 throughout, .ith $o i$dicatio$ to co$sumers that a pla$t has $ot met the sta$dard or has $ot bee$ tested i$ mo$ths& <& The poor impleme$tatio$ of the program has resulted i$ data that are riddled .ith problems& The poor data, a$d the Duestio$able a$alytical methods used to a$aly*e it, call i$to Duestio$ the validity of the age$cy's claims about the program's success at reduci$g the prevale$ce of !almonella i$ grou$d beef& Take$ together, these co$clusio$s illustrate that the S!"'s !almonella performa$ce sta$dard program remai$s a poor a$d u$prove$ replaceme$t for the loss of co$ti$uous gover$me$t i$spectio$ u$der H"CCP& The i$freDue$t testi$g rate a$d the age$cy's failure to act .he$ testi$g reveals that pla$t food safety systems are i$adeDuate, provide little i$ce$tive for grou$d beef producers to improve their systems&
PART FO&R: R"COMM"'DAT(O'S 0e recomme$d that the !almonella testi$g program be cha$ged by redesig$i$g the testi$g protocol& " more protective testi$g program .ould i$clude: K !aily testi$g& This elimi$ates .aiti$g periods bet.ee$ sets a$d /$ !epth Lerificatio$ revie.s& K Lolume,based testi$g, .hich .ould take i$to accou$t the amou$t of product se$t to market by differe$t si*e pla$ts& /$ other .ords, large pla$ts .ould be sub:ect to more sampli$g tha$ small a$d very small pla$ts& K Bover$me$t sampli$g for other pathoge$s, at a mi$imum, -. coli& K " .ar$i$g system to alert age$cy officials .he$ the sampli$g rate falls belo. daily, .hich .ould prompt a gover$me$t i$spector i$ the pla$t to report .hy daily sampli$g is $ot happe$i$g& These reports should be made public& K /mmediate $otificatio$ of compa$ies, i$spectors, a$d the public .he$ a pla$t fails a set by e%ceedi$g the allo.able $umber of failures& K " limit o$ the period of time i$ .hich pla$ts may release product .hile impleme$ti$g corrective actio$s& K Public release of $ames of pla$ts faili$g to meet the sta$dard& This should i$clude promi$e$tly displayi$g o$ the S!" .ebsite the results of faili$g sets alo$g .ith associated data such as dates for sample collectio$ a$d a$y corrective actio$s take$& K The prevale$ce rate that serves as the performa$ce sta$dard should get lo.er over time& K Publicatio$ of the testi$g protocol, .ith a detailed e%pla$atio$ of ho. sta$dards .ere chose$& K " more tra$spare$t e%pla$atio$ of the a$alysis of complia$ce a$d prevale$ce rates, a$d the use of co$siste$t a$alytical methods from report to report& "ll data should be i$cluded, .hich .ould elimi$ate the use of 8" sets&8 Ceports should address a$d correct for pote$tial biases& K " recalculatio$ of the complia$ce a$d prevale$ce data from the program to date, usi$g all data (.ithout 8" sets8)& To .hatever e%te$t possible, the age$cy should correct for a$y bias, a$d at a mi$imum, e%plai$ a$d Dua$tify its effect o$ results& K "s soo$ as possible, i$corporate rapid testi$g tech$ology i$to daily gover$me$t i$spectio$ activities& K Provide .histleblo.er protectio$ for i$dustry employees .hose disclosures are eve$ more sig$ifica$t i$ the abse$ce of reliable testi$g data& #cknowledgments The authors of this report .ould like to tha$k the follo.i$g for their research a$d editi$g assista$ce: +arti$ Ad.i$ "$derse$ Tom !evi$e Arica Hartma$ Tim Laias Peter Lurie Caitli$ +iller Natio$al 1oi$t Cou$cil of Food /$spectio$ Locals !o$$a Cose$baum -ob Sti% -ria$ 0olfma$&
Footnotes 2& 8Live from N+": +ura$o charts $e. fro$tier for food safety&8 !a$ +urphy& 6966976& http:99...&meati$gplace&com 6& 8Stateme$t for the Cecord by Stephe$ +& 5stroff, +&!&, "ssociate !irector for Apidemiologic Scie$ce, Natio$al Ce$ter for /$fectious !iseases, Ce$ters for !isease Co$trol a$d Preve$tio$, !epartme$t of Health a$d Huma$ Services, -efore the "griculture, Nutritio$ a$d Forestry Committee, &S& Se$ate&8 3967977& http:99...&cdc&gov9.ashi$gto$9legislative973676777&htm =& 8+oder$ +eat&8 Fro$tli$e& G924976& http:99...&pbs&org9.gbh9pages9fro$tli$e9sho.s9meat9etc9script&html G& 3 CFC =27&6>(b)& The other compo$e$t of the microbial testi$g program u$der H"CCP is a ge$eric A& coli sampli$g program co$ducted by the compa$y a$d audited by gover$me$t meat i$spectors& /$spectors are o$ly e$titled to see a summary chart of results a$d compa$ies are allo.ed to do multiple tests a$d o$ly report passi$g results& 3 CFC =27&6>(a)& >& 8/ssue Paper: Public Celease of Salmo$ella Testi$g Cesults&8 S!" Food Safety a$d /$spectio$ Service& 2692>93;& http:99...&fsis&usda&gov95"9haccp9issue72&htm <& 8Pathoge$ Ceductio$: Ha*ard "$alysis a$d Critical Co$trol Poi$t (H"CCP) SystemsI Fi$al Cule&8 S!" Food Safety a$d /$spectio$ Service& ;96>93<& http:99...&fsis&usda&gov9oa9fr9haccpNrule&htm (p& =44=>)& ;& H"CCP Fi$al Cule p& =44G4,3& 4& H"CCP Fi$al Cule p& =44G3& 3& 6;> F& =d G=6 (>th Cir& 6772) 27& Perso$al commu$icatio$& S!" Food Safety a$d /$spectio$ Service& >92;976 22& Perso$al commu$icatio$& S!" Food Safety a$d /$spectio$ Service& >92;976 26& 8S!" To Co$ti$ue Testi$g For Salmo$ella /$ +eat Pla$ts&8 S!" Ne.s Celease No& 76<;&72& 26924972& http:99...&usda&gov9$e.s9releases96772926976<;&htm 2= 8+ura$o does$'t embrace e%pa$sio$ of S!" authority8 Food Chemical Ne.s !aily& =92>976& Lol& G, No& 2;<& 2G& The age$cy should have sought out this level of e%pertise .hile it .as desig$i$g the program& 2>& The S!" categori*es .hat is do$e to meat as either slaughter or processi$g& 8Slaughter8 pla$ts .ork .ith products u$til they are cut do.$ to carcasses& "fter that poi$t, .hatever is do$e to the product is co$sidered processi$g& The age$cy's curre$t Salmo$ella regulatio$s apply at slaughter pla$ts for cattle, hogs, chicke$ a$d turkeys, a$d processi$g pla$ts that gri$d beef, chicke$ or turkey, but $ot to some other types of processed meats& 8-ecause Salmo$ella is more likely to be prese$t o$ ra., grou$d or commi$uted products tha$ o$ the carcasses from .hich they are derived, ra., grou$d, or commi$uted product ordi$arily .ill be the focus of Eage$cyF complia$ce testi$g i$ those establishme$ts that both slaughter a$d produce ra. grou$d product&8 H"CCP Fi$al Cule p& =44G4& 2<& "ccordi$g to the data .e received, the follo.i$g $umber of pla$ts for each product failed at least o$e 8sample set8 (a set co$sists of >= samples& See page ; for more e%pla$atio$ of the testi$g program protocol&): -roilers , G7I Co.s9-ulls , 4I Steers9Heifers , 2I Turkeys , >I +arket Hogs , 6GI Brou$d Turkey , =I Brou$d Chicke$ , 7I Brou$d -eef , <6& 2;& S!" a$aly*es grou$d beef samples ra$domly collected at federally i$spected productio$ facilities, retail stores, a$d import facilities, as .ell as samples collected by state perso$$el at state,i$spected pla$ts& The i$crease i$ the $umber of samples testi$g positive from all sources combi$ed rose from &7<H i$ 233; to $early 2H (&4<H) i$ 6777& 8+icrobiological Cesults for Ca. Brou$d -eef Products "$aly*ed for -scherichia coli /+01321&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95PHS9ecoltest9i$de%&htm 0hile much of this i$crease may be due to the adoptio$ of i$creasi$gly se$sitive testi$g methods, .e ca$$ot determi$e that this alo$e accou$ts for the i$crease& 24& 8Co$sumer appetite for beef remai$s stro$g, year,e$d data sho.s&8 -rya$ Salvage& 692=976& http:99...&meati$gplace&com 23& http:99...&fsis&usda&gov9ofo9faim9faimmai$&htm& 67& 0e .ere told that that all i$formatio$ give$ to us by a$y departme$t at the age$cy must first be cleared through the public affairs departme$t& 62& The age$cy ack$o.ledges that this could have happe$ed also bet.ee$ a pre, H"CCP testi$g phase a$d the begi$$i$g of H"CCP& 8Pathoge$ Ceductio$9H"CCP Salmo$ella Performa$ce Testi$g&8 S!" Food Safety a$d /$spectio$ Service& 3964934& http:99...&fsis&usda&gov95PHS9salmstd&htm 66& Perso$al commu$icatio$& S!" Aco$omic Cesearch Service& G92<976& 6=& 8+oder$ +eat8 Fro$tli$e& G924976& http:99...&pbs&org9.gbh9pages9fro$tli$e9sho.s9meat9etc9script&html 6G& 3 CFC =27&6>(b)(6) 6>& H"CCP Fi$al Cule p& =44G4& 6<& H"CCP Fi$al Cule p& =432;& 6;& 8!irective 27,722&2: A$forceme$t /$structio$s for the Salmo$ella Performa$ce Sta$dards&8 S!" Food Safety a$d /$spectio$ Service& 393934& http:99...&fsis&usda&gov95PP!A9rdad9FS/S!irectives927722,2&pdf ("ttachme$t 2)& 64& Lore$ La$ge, "ssista$t !eputy "dmi$istrator, 5ffice of Public Health a$d Scie$ce, Food Safety a$d /$spectio$ Service, prese$tatio$ at FS/S' 8Pathoge$ Ceductio$: " Scie$tific !ialogue&8 >9;976& 0ashi$gto$, !C& 63& H"CCP Fi$al Cule p& =44G;& =7& 3 CFC =27&6>(b)(2)& =2& 8!irective 27,722&2: A$forceme$t /$structio$s for the Salmo$ella Performa$ce Sta$dards&8 S!" Food Safety a$d /$spectio$ Service& 393934& http:99...&fsis&usda&gov95PP!A9rdad9FS/S!irectives927722,2&pdf (L/&!&G)& The le$gth of the .aiti$g period is sub:ect to age$cy discretio$ a$d may be affected by actio$s take$ by the pla$t& =6& 8Failure by the establishme$t Eto take appropriate corrective actio$sF, or failure to meet the sta$dard o$ the third co$secutive series of FS/S,co$ducted tests for that product & & &.ill cause FS/S to suspe$d i$spectio$ services&8 3CFC =27&6>(b)(=) (iii)& ==& 8+icrobiological Cesults for Ca. Brou$d -eef Products "$aly*ed for -scherichia coli /+01321&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95PHS9ecoltest9i$de%&htm =G& 8!irective 27,722&2: A$forceme$t /$structio$s for the Salmo$ella Performa$ce Sta$dards&8 S!" Food Safety a$d /$spectio$ Service& 393934& http:99...&fsis&usda&gov95PP!A9rdad9FS/S!irectives927722,2&pdf =>& "lthough .e have data for ;< complete sample sets at 6< large grou$d beef pla$ts, data for three of the first sets .as erro$eous because the e$d date of the set .as later tha$ the sampli$g dates for subseDue$t sets& !ata from a fe. small pla$ts had this same problem& /$ February 6776, age$cy officials co$firmed that these e$d dates must be erro$eous, but did $ot respo$d to repeated calls for correct sampli$g dates& =<& 0e asked S!"'s public affairs departme$t if a$y of the large pla$ts .ith delays lo$ger tha$ $i$e mo$ths had sig$ifica$t breaks i$ productio$ duri$g the time of the lapses i$ testi$g, a$d .ere told that the Food Safety a$d /$spectio$ Service does $ot keep track of this type of i$formatio$& 0e the$ asked gover$me$t i$spectors i$ the appropriate regio$s, .ho reported that they did $ot k$o. of a$y sig$ifica$t breaks at these pla$ts& =;& Cegulatio$s reDuire pla$ts to 8take immediate actio$ to meet the sta$dard,8 after the first failure, a$d to 8reassess its H"CCP pla$ for that product a$d take appropriate corrective actio$s,8 after the seco$d failure& 3 CFC =27&6> (b)(=)(/) a$d (ii) =4& 2>7,777 pou$ds9shift % 6 shifts9day % < days9.eek % 262 .eeks O 62;&4 millio$ pou$ds =3& The age$cy collected samples comparatively freDue$tly a$d co$siste$tly from +arch 2334 through 1u$e 2334& /t collected the last three samples i$ "pril, 2333 to complete the first Salmo$ella set& /t bega$ the testi$g for the seco$d set i$ 1uly 6777& G7& 8"$ 5utbreak 0aiti$g to Happe$: -eef,/$spectio$ Failures Let /$ a !eadly +icrobe&8 1oby 0arrick& G923972& 0ashi$gto$ Post& G2& Cemarks at Co$sumer Federatio$ of "merica's 8Natio$al Food Policy Co$fere$ce&8 G966976& 0ashi$gto$, !C& G6& 8!irective >>77&2: Co$ducti$g Targeted /$,!epth Lerificatio$ Cevie.s&8 S!" Food Safety a$d /$spectio$ Service& 27922972& http:99...&fsis&usda&gov95PP!A9rdad9FS/S!irectives9>>77&2&pdf G=& 8!irective >>77&2: Co$ducti$g Targeted /$,!epth Lerificatio$ Cevie.s&8 S!" Food Safety a$d /$spectio$ Service& 27922972& http:99...&fsis&usda&gov95PP!A9rdad9FS/S!irectives9>>77&2&pdf GG& "lthough there are fe. pla$ts that have had $umerous sets, it appears that the testi$g cycle le$gth may be i$creasi$g for later sets& H"CCP Fi$al Cule p&=44=; G>& !uri$g the 6; mo$th lapse, the age$cy collected eight samples& The age$cy completed G> samples for the first set by 1uly 4, 2334& To complete the set, it took three samples i$ November 2333, three samples i$ +ay 6777 a$d t.o samples i$ 1u$e 6777& /t started the seco$d set o$ 5ctober 22, 6777& G<& "dditio$ally, after a co$sumer discovered glass i$ grou$d beef from this pla$t, 27,777 pou$ds of the grou$d beef produced duri$g the test,free period .ere recalled& 8Cecall /$formatio$ Ce$ter&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95"9recalls9recNi$tr&htm G;& Had there i$stead bee$ i$visible pathoge$s i$ the grou$d beef, the co$sumer .ould have had $o .ay of k$o.i$g $ot to co$sume the product, perhaps resulti$g i$ serious co$seDue$ces& G4& !uri$g the $early 6G,mo$th lapse, the age$cy collected seve$ samples& The age$cy completed G; samples for the first set by "ugust G, 2334& To complete the set, it took three samples i$ November 2333, a$d three samples i$ November 2333& /t started the seco$d set o$ 1uly 23, 6777& G3& "ccordi$g to age$cy Puarterly A$forceme$t Ceports, the age$cy put the pla$t u$der suspe$sio$ o$ Friday 1u$e 2<, 2333& /t is possible, but age$cy reports do $ot specify, if this .as because the age$cy collected the si%th positive o$ 1u$e 27, 2333& The age$cy may i$stead have discovered a$d bee$ respo$di$g to a$other failure of a food safety co$trol& /$ a$y case, the i$terruptio$ .as short,lived because that suspe$sio$ .as put i$ abeya$ce by +o$day 1u$e 23, 2333, a$d the pla$t .as back i$ operatio$& >7& 2>7,777 pou$ds9shift % 6 shifts9day % < days9.eek % 6 pla$ts % G Duarter, pou$d hamburgers9pou$d O 2G&G millio$ Duarter,pou$d hamburgers& >2& H"CCP Fi$al Cule p& =44G3& >6& 8Pathoge$ Ceductio$9H"CCP Salmo$ella Performa$ce Testi$g&8 S!" Food Safety a$d /$spectio$ Service& 3964934& http:99...&fsis&usda&gov95PHS9salmstd&htm8Seco$d Progress Ceport o$ Salmo$ella Testi$g for Ca. +eat a$d Poultry Products&8 S!" Food Safety a$d /$spectio$ Service& 2962933& http:99...&fsis&usda&gov95"9backgrou$d9salmback&htm 85$e,@ear Progress Ceport o$ Salmo$ella Testi$g for Ca. +eat a$d Poultry Products&8 S!" Food Safety a$d /$spectio$ Service& =93933& http:99...&fsis&usda&gov95"9backgrou$d9salmtest=&htm 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products&8 S!" Food Safety a$d /$spectio$ Service& 27933& http:99...&fsis&usda&gov95"9backgrou$d9salmtestG&htm 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products&8 S!" Food Safety a$d /$spectio$ Service& =977& http:99...&fsis&usda&gov95"9backgrou$d9salmtest>&htm 8/$terim Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95PHS9haccp9salmra.&htm 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products, 2334, 6777&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov9ophs9haccp9salmdata6&htm 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products, 2334, 6772&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95PHS9haccp9salmGyear&htm >=& 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products, 2334,6772&8 p&2& >G& 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products&8 >>& 8FS/S Ceports Co$ti$ued !ecli$e 5f Salmo$ella&8 S!" Ne.s Celease& =962977& http:99...&fsis&usda&gov95"9$e.s967779salmrel6&htm ><& 8H"CCP /mpleme$tatio$: Salmo$ella Complia$ce Test Cesults, 1a$uary 6<, 2334 to 1a$uary 6G, 6777&8 S!" Food Safety a$d /$spectio$ Service& http:99...&fsis&usda&gov95PHS9haccp9salcomp&pdf >;& 8/$terim Progress Ceport o$ Salmo$ella Testi$g of Ca. meat a$d Poultry Products&8 >4& This .as determi$ed by usi$g the formula -9", .here " O the $umber of faili$g sets that follo.ed a passi$g set (6>)9the $umber of passi$g sets that had a subseDue$t set (=6<) a$d -O the $umber of faili$g sets that follo.ed a failed set (23)9the $umber of faili$g sets that had a subseDue$t set (;;) >3& 8S!" !ata Sho. a Ceductio$ of Salmo$ella i$ Ca. +eat a$d Poultry&8 S!" Ne.s Celease No& 72>G&76& G924976& http:99...&usda&gov9$e.s9releases9677697G972>G&htm <7& 66,673 of the ;G,;>4 samples (63&;H) used for the 2334,6772 aggregate prevale$ce figure (combi$i$g results from large, small a$d very small grou$d beef pla$ts) .ere from very small grou$d beef pla$ts& 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products, 2334,6772&8 <2& Lore$ La$ge, "ssista$t !eputy "dmi$istrator, 5ffice of Public Health a$d Scie$ce, Food Safety a$d /$spectio$ Service, prese$tatio$ at FS/S' 8Pathoge$ Ceductio$: " Scie$tific !ialogue&8 >9;976& 0ashi$gto$, !C& La$ge stated that very small pla$ts .ere $ot eve$ i$cluded i$ baseli$e testi$g because they are such a small perce$tage of productio$& <6& The very small grou$d beef pla$ts had a much lo.er prevale$ce rate (6&GH for 2334,6772) tha$ the large pla$ts (>&6H for 2334,6772)& 8Progress Ceport& & & 2334,6772&8 http:99...&fsis&usda&gov95PHS9haccp9salmGyear&htm& <=& Large pla$ts co$tributed =,3>G samples compared to the 66,673 samples co$tributed by the very small pla$ts& 8Progress Ceport o$ Salmo$ella Testi$g of Ca. +eat a$d Poultry Products, 2334,6772&8 <G& /$ co$trast, amo$g broiler pla$ts, it is the large pla$ts that co$siste$tly have lo.er prevale$ce rates& For the 2334,6772 aggregate figures, the prevale$ce rate for Salmo$ella .as much higher for the very small pla$ts (=G&;H) tha$ for small (2=&;H) or large (3&6H) pla$ts& /t is .orth $oti$g that, i$ broilers, .here the very small pla$ts did so poorly, they .ere $ot over,represe$ted by the perce$tage of samples from their class used for aggregate results , o$ly 2&GH of the samples for the 2334,6772 aggregate prevale$ce rate for broilers came from very small pla$ts& <>& http:99...&fsis&usda&gov9oa9co$gress9testNglickma$&htm <<& >4= samples .ere i$cluded from large pla$ts, .ith a prevale$ce rate of 3&2H& There .ere 4G6; total samples& ECopyright Q Privacy Stateme$t Q Terms a$d Co$ditio$sF