oversight

Food Safety and Quality: FDA Surveys Not Adequate to Demonstrate Safety of Milk Supply

Published by the Government Accountability Office on 1990-11-01.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

No\wIlt,t~r     l!I!IO
                                   FOOD SAFETY AND
                                   QUALITY
                                   FDA Surveys Not
                                   Adequate to
                                   Demonstrate Safety of
                                   Milk SUPPlY


                                                                              illllll~~lllll
                                                                               142662




                                                           RELEASED
                           RESTRICTED--Not       to be released outside the
                           General Accounting OlTlce unless epeclf’lcally
                           approved by the Office of Congressional
                           Relations.


GAO/H(        xn-9   1-N
                   united
                     states                                                              ,
GA!!!0             General Accounting Office
                   Washington, DE. 20548

                   Resources, Community,   and
                   Economic Development    Division

                   B-241106
                   November 1,199O
                   The Honorable Ted Weiss
                   Chairman, Human Resourcesand Intergovernmental
                     Relations Subcommittee
                   Committee on Government Operations
                   Houseof Representatives
                   Dear Mr. Chairman:

                   In your April 2,1990, letter, and in subsequentdiscussionswith your
                   office, you asked us to review the adequacy of the surveys the Food and
                   Drug Administration (FJ&Y)  conducted in 1988 and 1990 to determine the
                   presenceof selectedantibiotic drug residues in milk and whether the
                   information developedprovided a sufficient basis for FDA’spublic state-
                   ments on the safety of the milk supply.

                   FI~Ais responsible for assuring the safety of the billions of gallons of
                   milk produced in the United States each year, as well as numerous other
                   food, drug, and cosmetic products. m overseesthe nation’s milk supply
                   through a cooperative program with all states and the District of
                   Columbia (states). Generally, the states carry out the day-to-day over-
                   sight of the milk supply. FDAmonitors the overall cooperative program,
                   provides technical assistance,approves animal drugs for use in dairy
                   cows, and performs studies, takes samples,and does additional testing
                   when agency officials believe it is needed.

                   Concernedabout media reports that independent surveys had found a
                   variety of animal drugs (primarily antibiotics) contaminating the milk
                   supply, FDA conducted three efforts to determine the presenceof
                   selectedantibiotic drug residues in milk between 1988 and 1990. These
                   efforts were necessarybecause,except for penicillins, there is no routine
                   testing required to screenmilk for such drugs, many of which are not
                   approved for use in milk-producing dairy cows (dairy cows). FDA stated
                   that the results of these three surveys confirmed its belief that the
                   nation’s milk supply was safe.

                   FDAstatements that the nation’s milk supply was not contaminated with
Results in Brief   unsafe animal drug residues cannot be supported becauselimitations in
          Y        the survey methodologies precluded any overall conclusions.Specifi-
                   tally, the surveys were not statistically valid and present, at best,
                   “snapshots in time” of a small number of milk samplestested for the


                   Page 1                                    GAO/RCED91-26Drug Jtesiduesin Milk
             E241106




             presenceof a small number of drug residues.However, collectively,
             becausethe surveys show instancesof drug residues in milk, they sug-
             gest a need for more thorough examination by FDAto identify the types
             and amounts of animal drug residuesthat may be contaminating milk.
             Even if the surveys had been statistically valid, the results would still be
             of limited use becauseFDAdoesnot have test methods to detect and con-
             firm many drugs believed to be used in dairy cows. Generally, compa-
             nies submitting drugs to rn~ for approval for use in food-producing
             animals must develop tests to identify any residues left by the drug in
             edible tissues or milk. However, many of the drugs believed to be used
             by the dairy industry were not submitted to m for use in dairy cows,
             and the tests to detect their residues in milk have not been developed.In
             other cases,the test methods FDAhas are not sensitive enough to confirm
             the presenceof drug residues at the health concern levels set for human
             consumption.
             Our review, which was limited to FDA’smilk surveys conducted between
             1988 and 1990, raised other questions about the adequacy of routine
             monitoring of the milk supply by FDAand cooperating state agencies,
             m’s “extra-label use” policy that permits the use of drugs not specifi-
             cally approved for dairy cows, and the setting of health concern levels
             for unapproved drugs.

             Under the Federal Food, Drug, and CosmeticAct (FFDCA),     as amended,
Background   FVA,part of the US. Department of Health and Human Services(HHS), is
             responsible for ensuring the safety of the nation’s milk supply. The FIN’S
             milk safety program is a collaborative federal/state effort that dates
             back to the mid-1920s. Most milk is produced and marketed under the
             Grade A Pasteurized Milk Ordinance (the Milk Ordinance). The Milk
             Ordinance is recognizedby public health agencies,the milk industry,
             and many others as a national standard for milk sanitation.
             The only official test for detecting animal drugs in milk, under the Milk
             Ordinance, is the Bacillus Stearothermophilus Disk Assay test (disk
             assay). While the disk assay effectively detects the residues of several
             drugs in the penicillin family, it is much less effective in detecting many
             of the other drugs now being used by the dairy industry. For instance,
             the disk assay detects sulfa drugs at levels of 16 parts-per-million (ppm)
             or higher-l,600 times the 10 parts-per-billion (ppb)concern level set by
             a for milk. Somestates and industry groups supplement the disk assay
             with other tests that are more sensitive to sulfa and other drugs. F’I)A


             Page 2                                     GAO/ECED-Bl-26Drug Beeiduar in Milk
                                                                                        .
E!441106




itself doesnot routinely test milk samplesfor drug rosiduos but relies
instead on the states and industry for such testing. However, according
to FDAMilk Safety Branch officials, FQAdoesnot routinely receive state
or industry test results.
FDAis also responsible for determining whether new animal drugs, such
as antibiotics, are safe and effective for those animals and whether the
food products (including milk) from treated animals will be safe for
human consumption. Generally, new animal drugs may be legally mar-
keted in the United States only if FI%has determined that they are safe
and effective and has established tolerances for their intended uses.
A tolerance defines the amount of residues of a new animal drug from
treated animals that is demonstrated to be safe in the human diet.1F+W
also sets withdrawal periods for approved drugs during which time
meat or milk products from the treated animal cannot be marketed. The
intent is to allow the drug to be purged from the animal’s system so that
any residues are below the tolerance level (see app. I for more informa-
tion on how FDAestablishestolerances for drug residues in milk).
Residuesof drugs, including antibiotics, can occur in milk as the result
of legal or illegal use of drugs. FDAhas approved 63 drugs for use in or
on dairy cows, including 20 antibiotic drugs. In addition, FDAhas estab-
lished tolerances in milk for 21 of these 63 approved drugs. Generally,
an illegal use occurs when a drug residue is found that exceedsits toler-
ance level, when misuse of a drug approved for use in dairy cows results
in residues in milk for which no tolerance has been established, or when
residues of a drug not approved for use in dairy cows results in residues
in milk.
Actual or intended use of a new animal drug in a food-producing animal
in a manner inconsistent with the approved labeling can result in FDA
taking regulatory action against the veterinarian, producer, or other
personsinvolved. However, FDAhas established guidelines for veterinar-
ians to treat food-producing animals with drugs in an unapproved
manner, if suffering and/or death would result from not treating the
affected animal.


%?dncemostof the drugs FDAtestedfor in its surveys are not approved for usein dairy cowsand
have no official tolerancelevels,FDAset unofficial “concern1eveWfor usein the surveys.Generally,
theseconcernlevels were estimatedon the bssisof tolerancelevels establishedfor the drugs in other
animal speciesand tissues.



Page 8                                                 GAO/ltCED&24I Drug Iteddn~ in Milk
   ,                E241106




                    In establishing what is known as the extra-label use policy, FDA stated
                    that it would ordinarily refrain from taking regulatory action against
                    licensed veterinarians for using or prescribing any drugs they could
                    legally obtain provided certain criteria are met. FW’Spolicy doesnot
                    permit non-veterinarians (e*g.,dairymen) to treat food-producing ani-
                    mals with drugs in an unapproved manner. In addition, FDA has declared
                    that certain drugs cannot be used under the extra-label use policy
                    becauseof public health concerns.

                    Between 1988 and 1990, FDAconducted three efforts to determine the
FDA Milk Survey     presenceof selectedantibiotic drug residues in milk in responseto
Methodologies and   reports from the media and others that drugs were contaminating the
Results             milk supply. In March 1988, FDA tested a total of 49 retail milk samples
                    from 10 cities in the United States to determine the presenceof
                    sulfamethazine (SMZ), an antibiotic sulfa drug, in milk. SMZ is not
                    approved for use in dairy cows, and its residues in milk may pose a risk
                    for individuals allergic to sulfa-baseddrugs. SMZ is also a suspectedcar-
                    cinogen (cancer-causingagent). FDAfound that 73 percent of the samples
                    tested (36 of 49) contained SMZ levels ranging from 0.8 ppb to 40.3 ppb.
                    Five of the 36 samplescontained SMZ residues above 10 ppb, FDA’sunoffi-
                    cial concern level for SMZ at that time.
                    Prompted by the 1988 survey results, FDAtook several steps intended to
                    eliminate SMZ residues in milk, including an educational campaign aimed
                    at dairy farmers. FDA coordinated this educational effort with the
                    National Conferenceon Interstate Milk Shipments (the National Confer-
                    ence).2In October 1988, the National Conferencesent a questionnaire to
                    all state milk regulatory laboratories asking for data on raw milk sam-
                    ples tested for SMZ from May to September 1988 to assessthe effective-
                    nessof FDA’S educational program.
                    FDA’sanalysis of this follow-up survey indicated that 6 percent of
                    reported samplestested (247 out of 4,887) contained SMZ residues and 1
                    percent (64 out of 4,887) contained SMZ residues above 10 ppb.Basedon
                    these results, FDAconcluded that SMZuse in dairy cows had decreased
                    significantly since its 1988 survey and the SMZ problem had been
                    resolved.

                    ?he National Conferenceis a voluntary organizationof federal and statehealth and agricultural
                    officials and the dairy industry that, alongwith FDA,overseesa cooperative,federal-stateprogram
                    (the Interstate Milk ShippersProgram)to ensurethe sanitary quality of milk and milk products
                    shippedinterstate.



                    Page 4                                                  GAO/ICED-91-26 Drug Residue6in Milk
E!u1105




FDA’Sconclusion was subsequently questioned in December1989, when
the Wall Street Journal (the Journal) reported the results of two surveys
of animal drug residues in milk, one sponsoredby the Journal and the
other sponsoredby the Center for Sciencein the Public Interest (CSPI), a
consumer food safety and nutrition organization, The Journal survey
found that 38 percent of 60 retail milk samplescontained antibiotic resi-
dues, possibly including SMZand other unapproved drugs. The CSPI
survey foundthat 20 percent of 20 retail milk samplescollected in the
Washington, D.C., area contained sulfa drugs, again possibly including
SMZand other unapproved drugs. Both surveys used an analytical
method called “Charm II.” This method is considereda screeningtest
becauseit reportedly detects the presenceof seven classesof antibiotic
drug residues,but generally cannot identify individual drugs within
these classes.

In responseto media reports of contaminated milk, FDAconducted
another survey in 1990 to test the reliability of the independent surveys
and to determine for itself the presenceof animal drug residues in milk.
FDAcollected 70 retail milk product samples3-5 each from 14 cities
sampled in its 1988 survey and/or the Journal survey. FDAalso used the
Charm II, as well ‘asother methods, to test the milk samples.
FDA, as a matter of policy, requires further, more specific tests of all
positive screeningtest results such as those obtained from Charm II to
conclusively identify and confirm the presenceof the specific drug resi-
dues present. These more specific tests include high pressure liquid
chromatography (HPLC), thin layer chromatography, gas chromatog-
raphy, and mass spectrometry test methods. FDAofficials consider the
more costly and difficult mass spectrometry testing to be the most reli-
able confirmation method for identifying specific drug residues for
enforcement purposes.
FDA’sresults using the Charm II test were similar to the results from the
Journal and CSPIsurveys in that the presenceof antibiotic drug residues
was indicated in many of the samplestested. In addition, the results of
HPLc testing at FDA’S Beltsville laboratory indicated that almost 86 per-
cent of the samplestested (60 of 70) contained sulfa drug residues and
that 11 of the samples(16 percent) had residues above the concern
levels for the drugs analyzed. However, upon subsequentconfirmation
testing with mass spectrometry methods, FDAdid not find any of the

3FJMcollected2 identical containersfrom eachof 6 storesin 14 different cities and sentoneof each
to its Beltsville and Philadelphia laboratories,respectively.



Page 5                                                  GAO/WED-91-26Drug Residuesin Milk
   .
                        5&(llW




                        antibiotics it tested for above established health concern levels, or above
                        the level of detection sensitivity of the methods used.
                        FR.% initially reported, in a February 61990, press release,that it could
                        not confirm the presenceof any antibiotics in the milk samplestested in
                        the 1990 survey. However, FDA’spress releasewas premature because
                        the agency had not completed its analysis of the milk samplesat the
                        time it was issued.Upon subsequenttesting, FDAconfirmed the presence
                        of SMZat levels less than 6 ppb in three milk samplestested. FDAlater
                        modified its presentation of the 1990 survey results to report this
                        finding.


                        ~~4’sefforts and the independent surveys were basedon a limited
FDA’s Survey Results    number of milk samplesthat were not selectedin a manner that would
Not Representative of   allow drawing statistically valid conclusionsabout the overall milk
the Nation’s Milk       supply. At best, m’s efforts and the independent surveys were snap-
                        shots in time of a small number of milk samplestested for the presence
SUPPlY                  of a small number of drug residues.

                        With specific regard to FIN’Sinitial 1988 survey and its 1990 effort, a
                        total of 119 milk sampleswere tested-a very small percentageof the
                        overall milk supply. In addition, the milk products and sampleswere not
                        randomly selectedfor these surveys.
                        FIIA’S1988 follow-up to its initial 1988 survey is not comparable and
                        doesnot show that the SMZproblem initially found in its survey of 10
                        cities was corrected. First, the 23 states that respondedto the survey
                        that tested milk samplesproduce only about 66 percent of the nation’s
                        milk supply annually. More importantly, FDAcould not provide docu-
                        mentation to show how the responding states had sampled milk prod-
                        ucts. If the testing states did not use statistical methods to obtain sample
                        data, then the resulting state and national data are highly suspect.FDA
                        also did not know whether the analytical methods and calibration stan-
                        dards used by the testing states to detect SMZresidues were similar or
                        whether the methods used were all capable of detecting SMZ residues at
                        the 10 ppb health concern level used by FDAat that time.
                        We cannot state definitely that the SMZresults FIN developed from its
                        1988 follow-up effort are flawed. However, with a sample representing
                        only 66 percent of the milk supply, no assurancesthat state samples
                        were statistically drawn, and questions regarding the comparability of



                        Page 6                                     GAO/ltCJD91-26 Drug Residuesin Milk
                    E!241105




                    the testing performed, for the FDAestimates to be correct would likely be
                    coincidence.

                    In FDA’S1988 and 1990 survey efforts, only a limited number of drugs
Limited Number of   were tested in comparison to the total nymber of animal drugs that
Drugs Tested        might have been present in milk products. Approximately 78 drugs,
                    approved and unapproved, are believed to be used in the dairy industry.
                    During its surveys, FDAdid not test milk samplesto determine generally
                    whether many of the 63 drugs approved for use in or on dairy cows
                    were present, or for many of the approximately 26 drugs not approved
                    for use in dairy cows but believed to be used in the dairy industry.

                    In its 1988 survey and subsequentfollow-up effort, FDAonly tested for
                    or collected information on one drug-SMZ. Also, while FDAused the
                    Charm II test to screenthe 1990 survey samples for the presenceof
                    sevenclassesof antibiotics, FDAonly had methods available to confirm
                    the presenceof six of the antibiotics included in its survey.
                    FDAhas also observedthat drugs used to treat dairy cows are to some
                    extent chosenbecausetesting is not being performed to detect them, or
                    the drugs cannot be detected by available test methods. Consequently,
                    testing for a limited number of drug residues in sampled milk products
                    doesnot provide information on the presenceof other drugs that might
                    be in milk and may not provide sufficient information to support broad
                    conclusionson the safety of the milk supply.

                    m doesnot know what additional test methods states use to sample
SomeDrugs Not       milk or whether such tests can detect somedrugs at their concern level.
Deteckd at Their    For instance, regarding the 1988 follow-up survey, FDAdid not have
Concern Level       information on whether the 23 states that reported test results data
                    used test methods sensitive enough to detect SMZresidues at the then 10
                    ppb concern level. Somestates may have used only the disk assay which
                    is the only official screeningtest for detecting antibiotic drug residues in
                    milk under the Milk Ordinance. While the disk assay is effective for
                    detecting penicillins, it cannot detect many other antibiotics (including
                    SMZ)at the concern levels established by FN Although FDAofficials
                    believe that somestates had other test methods available to detect SMZ
                    residues, they do not know what methods the states used. If the
                    reported results were basedon the official disk assay method alone,
                    then the results might have understated the number of samplescon-
                    taining SMZabove the FLNconcern level.


                    Page7                                       GAO/RCED-Bl-26Ihag lteoiduer in MU
    .
                                  B!241106




                                  In addition, someof the analytical methods used in FDA’s1990 survey
                                  efforts were unable to detect and/or confirm somedrug residues at con-
                                  cern levels. For example, FDAdiscounted Charm II screeningtest results
                                  which indicated the presenceof novobiocin residues in five of the sam-
                                  ples tested on the basis of another screeningtest method that was not as
                                  sensitive as the detection capability claimed for the Charm II. Further-
                                  more, this secondscreeningmethod could not detect novobiocin at the
                                  health concern level established by FDA.
                                  FIX also requires that survey screeningtest results indicating the pres-
                                  enceof illegal or unapproved drug residues be confirmed using mass
                                  spectrometry testing, to facilitate its ability to take regulatory action.
                                  However, in its 1990 survey, FDAonly had mass spectrometry methods
                                  for six drugs, and three of these methods were incapable of confirming
                                  the presenceof their drugs at the health concern levels established. Spe-
                                  cifically, three of the six drugs for which FDAhad confirmatory tests
                                  were chlortetracycline, oxytetracycline, and tetracycline.
                                  m-established concern levels for these three antibiotics were/are 30,
                                  30, and 80 ppb, respectively. However, FDA's mass spectrometry methods
                                  could only confirm the presenceof these drugs at levels of 100 ppbor
                                  more. Thus, although FDAreported that it could not confirm the presence
                                  of any tetracycline drugs, the drugs may still have been present in
                                  amounts exceedingthe concern level, but below the level of the con-
                                  firming test’s sensitivity.

                                  Additional questions have been raised about FDA's handling of the milk
                                  samples and use of analytical procedures in its 1990 survey. These and
                                  other limitations are discussedfurther in appendix II.

                                  During this review, several related issueswere identified that we believe
Other Concerns                    merit further study. However, they were beyond the scopeand time
Relating to Drug                  available for completing this work. Specific issuesinclude the (1) ade-
Residuesin Milk                   quacy of routine drug residue screeningin milk by m and the states,
                                  (2) impact of FDA’Sextra-label use policy as it relates to dairy cows, and
                                  (3) setting of unofficial concern levels for unapproved drugs basedon
                                  limited data.


Routine Drug Residue      Neither FDAnor the states routinely screenmilk for many of the drugs,
Testing May Be”Inadequate approved  or unapproved, that might be used by the dairy industry.
                          With regard to antibiotic drugs, the only screeningtest officially sanc-
                                  tioned by the Milk Ordinance is effective primarily for detecting the


                                  Page 8                                    GAO/RCEDBl-26 Dnqj Residuesin Milk
                                                                                                  ,
                            B-241106




                            penicillin family of drugs, not the sulfa, tetracycline, or other drug fami-
                            lies. FDAis also aware that the choice of unapproved drug used is often
                            guided by the fact that regulators cannot detect or are not checking for
                            that drug. For these reasonsFDAhad to undertake the surveys discussed
                            in this report.
                            Somestates, veterinarians, and industry organizations, concernedthat
                            the Milk Ordinance’s official screeningtest for antibiotic drug residues is
                            not effective in detecting drugs other than penicillins, have asked FDAto
                            either develop reliable and inexpensive methods for their use or offi-
                            cially sanction someof the commercial tests that somestates already
                            use such as the Charm II method. However, according to the Deputy
                            Director, Center for Veterinary Medicine, m doesnot have the legisla-
                            tive authority to approve or sanction screeningtests that are not sub-
                            mitted as part of a sponsor’sapplication for a new animal drug
                            approval.
                            According to FDAattorneys, screeningtest kits by themselvesare consid-
                            ered animal devicesunder FFLXAand, as such, are not subject to pre-
                            market review and approval by FDA.However, such kits are subject to
                            post-marketing controls under the law. Thus, for example, the labeling
                            of screeningtest kits must be truthful, accurate, non-misleading, and
                            must bear adequate directions for use. FDAdoeshave the authority to
                            evaluate commercially available screeningtest kits and publish the
                            results of such evaluations.
                            FDAplans to exercise its evaluation authority in order to assist the states
                            and the dairy industry in determining/obtaining effective and reliable
                            analytical methods for detecting animal drug residues in milk. Starting
                            with sulfa drugs, FDAintends to collect information on commercially
                            available screeningtests, conduct limited evaluations of such tests-
                            including their capabilities to detect residues at or above the tolerance
                            or other levels established by FDA,and make the results of these evalua-
                            tions publicly available.


Extra-Label Drug Use        IQA’Sefforts to overseethe safety of the nation’s milk supply are com-
Makes Oversight Difficult   plicated by the use of drugs not specifically approved for use in dairy
                            cows and the extra-label use policy that allows it. In brief, this policy
                            allows veterinarians to use drugs in an unapproved manner on animals
           i                if that animal’s life is in danger, and no other effective approved drugs
                            are available.



                            Page 9                                      GAO/WED-91-26Drug Residuesin Milk
                          &241106




                          No illegal drug residues are supposedto result from the extra-label use
                          of drugs in dairy cows. However, neither FDAnor drug companieshave
                          performed the studies necessaryto establish the dosagelevels and with-
                          drawal periods neededto ensure that illegal/unsafe residue levels do not
                          result in milk from extra-label uses.The lack of such scientific data
                          makes it difficult for veterinarians to determine and prescribe dosages
                          and withdrawal times that will assureillegal residues do not occur.
                          Another concern is that someof these drugs are available to the layman
                          “over-the-counter,” As such, the veterinarian/client relationship on
                          which the extra-label drug use policy is basedoften may not exist.
                          FDAalso has not determined what “safe” milk residue levels are for
                          many of the drugs used in extra-label dairy applications. Instead, FDA
                          has set unofficial concern levels for a limited number of these drugs,
                          primarily for the 1990 survey. FDA’sgeneral practice is to not take regu-
                          latory action when residues are found at levels below the concern level.
                          Many of the residues detected in FW’S1988 and 1990 milk surveys were
                          from drugs not approved for use in dairy cows. Becauseof FDA’spolicy
                          that allows the extra-label use of drugs by veterinarians and its general
                          practice not to take action when drug residues are below the concern
                          level, it is difficult to determine whether residues found in milk result
                          from the improper use of drugs by veterinarians under the extra-label
                          use policy or from illegal use by non-veterinarians.


The Adequacy of Concern   Many of the drugs tested for in FDA’S1990 survey are not approved for
Levels Is Questionable    use in dairy cows and there are no official tolerance levels established
                          for their presencein milk. As a result, FIX conducted a health risk
                          assessmentto determine unofficial drug residue concern levels for
                          selecteddrugs targeted in its survey efforts. m estimated the concern
                          levels for residues in milk on the basis of tolerances established for the
                          drugs in other animal speciesor tissues or on other data. Questionsexist
                          about (1) the reliability of setting levels in this manner, (2) how factors
                          such as the aggregateand synergistic health effects of the drugs were
                          handled, and (3) whether drug metabolites,* both individually and those
                          of closely related drugs, should be included in assessinghealth effects.


                          *Drug compoundsadministeredto food-producinganimalscan form or be brokendown into new
                          substance8(metaboliteaand degradationproductsof the compound)by the animal’sbiologicalsy%
                          tans, which can poset.oxic&gical concernsof their own. Therefore,the total residueof a drug pro-
                          posedfor we in food-producinganimalsconsIstaof the parent drug and 1t.ametaboliteaand any other
                          substanceformedin or on food becauseof the useof the parent compound.



                          Page 10                                                GAO/RCJiXWl-28Drue Residueain Milk
              E241106




              FDAregards allergic responsesto penicillin, tetracycline, and/or sulfa
              drugs as a serious public health concern.Studies conducted by FDA’s
              National Center for Toxicological Researchalso indicate that moderate-
              to-high dosesof the sulfa drug SMZ causethyroid cancer in somelabora-
              tory animals. In addition, FDAdata suggestthat if a structural feature in
              one compound is found to causecancer,the presenceof that samestruc-
              tural feature in other compoundsgreatly increasesthe probability that
              they too can causecancer.All of the sulfa drugs have very similar struc-
              tures. According to FDAdata, someof these drugs exhibit toxic effects
              similar to those produced by SMZ.
              In setting concern levels for the 1990 milk survey, however, FM allowed
              residues of up to 10 ppb for each sulfa drug and did not consider the
              aggregateor synergistic effect of these drugs. The results of FLW’S HPLC
              screeningtests for the 1990 survey indicated that 46 percent (32 out of
              70) of the samplestested contained more than one sulfa drug residue.
              FDAdata also indicate that drug metabolite residues in food are an
              important consideration. FLMstates that the importance of any metabo-
              lite in terms of its level, persistence,structure, relationship to the parent
              drug, and the anticipated human exposure must be consideredin
              deciding about the need for separate toxicity testing. According to FM,
              drug metabolites are likely to present health risks which may be as
              important as residues from the parent drug becauseof their amount,
              persistence,or potential for toxicity. For example, according to FM data,
              the SMZ metabolite levels that develop in pork are 3 to 10 times higher
              than the detectable residue level of the parent drug.

              FDAsuspectsthat metabolites may also be present in milk when drug
              residues are found and is currently attempting to determine the amount
              of SMZ metabolites that may be “hidden” in milk. According to FDAdata,
              several of the metabolites that develop from this drug in milk may pre-
              sent carcinogenicrisks similar to those associatedwith SMZ itself. How-
              ever, metabolites could not be detected by the analytical methods FDA
              used in the milk surveys, and, as a result, the metabolite levels that
              might be found in milk were not considered.

              FDA’sefforts to determine the presenceof animal drug residues in milk,
Conclusions   as well as the independent surveys, were not statistically designedto be
         Y
              representative of the nation’s milk supply. Thus, the surveys do not pro-
              vide an adequate basis for the statements made by FDAregarding the
              nation’s milk supply. However, collectively, becausethe surveys show


              Page 11                                     GAO/ltCEWl-28 Drug Residuesln Milk
  1



                  B-241105




                  instancesof drug residues in milk, they suggestthe need for more thor-
                  ough examination to identify the types and amounts of animal drug resi-
                  dues that may be present in milk.
                  Furthermore, if FDA had designedand undertaken a statistically valid
                  random sample of milk products to test for drug residues that was rep-
                  resentative of the nation’s milk supply-a difficult and costly
                  endeavor-the results would still be of limited value becauseFDAlacks
                  test methods to detect and confirm most of the drugs believed to be used
                  in dairy cows, and someof FDA’S test methods cannot detect drug resi-
                  dues at the concern level set for human consumption.
                  The only screeningmethod sanctioned by the Milk Ordinance for antibi-
                  otic drug residues in milk, by itself, will not effectively detect most
                  drugs currently used by the dairy industry. Somestates are supple-
                  menting the Milk Ordinance’ssanctioned screeningtest with other com-
                  mercially available methods that are reportedly capable of detecting
                  more drug types.
                  Although FDA may lack sufficient legislative authority to officially sanc-
                  tion screeningtest methods that are not part of an application for new
                  drug approval, FDA can exercise its authority to evaluate commercially
                  available screeningtest kits and share the results of its evaluations with
                  the states, industry, and others. Beginning with sulfa drugs, FIIAplans to
                  assist the states and others by conducting these evaluations and pub-
                  licizing the results.
                  FDAhas also begun to develop new test methods capable of detecting and
                  confirming specific drugs at established concern levels. However, there
                  are many additional drugs, approved and unapproved, for which ade-
                  quate tests do not exist. These and other limitations in testing capability
                  detract from FDA’sability to ensure the safety of the milk supply.
                  Finally, FDA’sregulatory efforts to ensure a safe milk supply are made
                  more difficult by the extra-label use of drugs in an unapproved manner
                  in dairy cows and by questions regarding the basis for setting unofficial
                  concern levels and how to treat factors like the cumulative effect of
                  closely related drugs and drug metabolites.

                  Becauseof insufficient data to fully addressmany of the issuesdis-
Recommendations   cussedin this report, and recognizing that it is unlikely that FDA could



                  Page 12                                    GAO/RCED-91-26Drug Residuesln Milk
    E241106




    devote a large amount of its limited resourcesto this one issue,we rec-
    ommend that the Secretary, HHS,direct the Commissioner,FDA,to take
    several incremental actions to provide greater assurancethat the milk
    supply is safe.
    First, FDAshould,develop more complete information on the incidence of
    drug residues in milk. FDAshould begin by asking the states under their
    cooperative agreementsand the dairy industry to routinely provide
    them with the results of their screeningtests for drug residues in milk,
    as well as information regarding their sampling plan and the types and
    sensitivities of test methods employed.
    Second,to further assist state regulatory efforts, FDAshould work with
    the states to evaluate commercially available screeningtests and
    encouragethat those found effective for sulfa, tetracycline, and other
    drugs, be included in the Milk Ordinance as a supplement to the disk
    assay,which is primarily effective only for penicillins. If FDAdetermines
    that it needsadditional legislative authority to approve screeningtests
    apart from new drug applications, then it should seek such authority
    from the Congress.
    Third, m should prioritize and expedite its current efforts to develop
    and evaluate new screeningand confirmatory test methods for animal
    drug residues in milk, possibly according to the health risks they per-
    ceive to be associatedwith the individual drugs involved.

    Fourth, FDAshould work closely with the states to confirm, possibly on a
    random basis, the types and amounts of drug residues found in state
    screeningsamples.If this information and confirmatory testing indicates
    that potential problems exist, FDAshould work with the states to further
    expand testing.
    Last, if the additional information developed from increasedscreening
    and confirmatory testing indicates that widespread problems exist from
    the misuse of drugs approved and/or unapproved for use in dairy cows,
    m should reassessthe appropriateness of its policies on extra-label
    drug use and its use of concern levels as a trigger for regulatory action.


w   Our work was conducted from April to September 1990 at FDAheadquar-
    ters and field locations in the Washington, D.C., metropolitan area. (Fur-
    ther details on our objectives, scope,and methodology are provided in
    app. III.)


    Page 13                                    GAO/RCED-91-26Drug Residuesin Milk
     It-B11106




     We discussedthe information in this report with officials in FLX’SCenter
     for Food Safety and Applied Nutrition, Center for Veterinary Medicine,
     and Office of Regulatory Affairs. Where appropriate, somechanges
     have been made basedon the discussionto further clarify the informa-
     tion presented. However, as requested by your office, we did not obtain
     official agency comments on a draft of this report.
     As arranged with your office, unless you publicly announceits contents
     earlier, we will make no further distribution of this report until 30 days
     after the date of this letter. At that time we will send copiesto the Sec-
     retary, HHS;the Commissioner,m; interested congressionalcommittees;
     and other interested parties upon request.
     This review was conducted under the direction of John W. Harman,
     Director, Food and Agriculture Issues,who may be reached at (202) 276
     6138. Other major contributors are listed in appendix IV.

     Sincerely yours,




/J   J. Dexter Peach
     Assistant Comptroller General




     Page14                                    GAOjIKXXMl-28DrugResiduesin Milk
Page 16   GAO/BcEIM)l-26 Drug lteddriea in MilL
contents


Letter                                                                                         1

Appendix I
Federal   Regulation
                  of    Tolerance Setting Process
                        Monitoring Milk Safety
                                                                                             18
                                                                                             18
                                                                                             24
Drug Residuesin the
Milk Supply
Appendix II                                                                                  26
Results and             1988 Survey
                        1988 Follow-Up Survey
                                                                                             26
                                                                                             26
Limitations of FDA’s    1990 Survey                                                          28
Efforts to Determine
the Presenceof
Antibiotic Drug
Residuesin the
Nation’s Milk Supply
Appendix III                                                                                 39
Objectives, Scope,and
Methodology
Appendix IV                                                                                  40
Major Contributors to
This Report
Tables                  Table II. 1: FDA ScreeningResults for Drug Family                    31
                            Residuesin the 1990 Milk Survey Using the Charm II
                            Test
                        Table 11.2:Initial FDA ScreeningResults for Sulfa Drug               32
                            Residuesin the 1990 Milk Survey Using HPLC
                        Table 11,3:Tolerances/Levels of Concern for Selected                 36
                            Antibiotic Drug Residuesin Milk and the Level of
                            Detection for the Methods FDA Used in Its 1990
                            Survey (In Parts-Per-Billion)




                        Page 16                                GAO/RCED-Sl-28m   Realdueain Milk
Contenta                                                           -.




Abbreviations

ADI        acceptabledaily intake
CDC        Centers for DiseaseControl
CFSAN      Center for Food Safety and Applied Nutrition, FDA
CSPI       Center for Sciencein the Public Interest
CVM        Center for Veterinary Medicine, FDA
FDA        Food and Drug Administration
           Federal Food, Drug, and CosmeticAct
GAO        General Accounting Office
GAP        Government Accountability Project
HHS        U.S. Department of Health and Human Services
HPLC       high pressure liquid chromatography
ppb        parts-per-billion
wm         parts-per-million
SMZ        sulfamethazine


Page17                                   GAO/RCED-91-26Drug Residuesin Ml&
Appendix I *

Federal Regulationof                Drug             Residuesin the
Milk Supply

                    Under the Federal Food, Drug, and CosmeticAct (FFDCA), as amended
                    (21 U.S.C.301 et seq.),the Food and Drug Administration @DA)is
                    responsible for ensuring the safety and purity of the nation’s milk
                    supply. In addition, FDA is required to determine whether new animal
                    drugs for use in food-producing animals,1such as antibiotics for use in
                    milk-producing dairy cows (dairy cows), are safe and effective for those
                    animals and whether the edible products derived from treated animals,
                    such as milk, will be safe for human consumption. Under FFDCA and FDA
                    policy, food items containing unapproved animal drug residues that may
                    be harmful to humans are consideredto be adulterated and subject to
                    regulatory action.

                    Generally, a new animal drug may be legally marketed in the United
Tolerance Setting   States only if FDA has determined that it is safe and effective and has
Process             established tolerances for its intended uses.A tolerance is a’legally
                    binding limit that defines the amount of residuesof a new animal drug
                    in edible tissue (or other edible products such as milk) from treated ani-
                    mals that is demonstrated to be safe in the human diet.

                    FM usestoxicology and residue data to assesspossible health risks of an
                    animal drug residue in milk and determine the tolerance level that will
                    protect the public health within a practical certainty. The risk of a drug
                    residue dependson both the toxicity of animal drug residues(i.e., their
                    potential to causeadversehealth effects in humans) and potential
                    human exposure to residues in the diet.

                    Sponsorsof new animal drugs must submit data to FDA that substanti-
                    ates the safety and effectiveness of the proposed drug. The data must be
                    specific for each use and speciesof animal for which the drug is
                    intended and, for food-producing animals, must include evidence
                    showing the safety of residues of the drug (including metabolites2) and
                    acceptablemethods for recovering and measuring such residues from
                    edible products.



                    ‘Under FFDCA,new drugs are drugsthat are not generallyrecognizedby qualified experts assafe
                    and effective for their labeleduses.
                    ‘Drug compoundsadministeredto food-producinganimalscan form or breakdown into new sub-
                    stances(metabolitesand degradationproductsof the compound)by the animal’sbiologicalsystems,
                    which can posetoxicologicalconcernsof their own. Therefore,the t&al residueof a drug proposed
                    for usein food-producinganimalsconsistsof the parent drug and its metabolitesand any other sub
                    stanceformedin or on food becauseof the useof the parent compound.



                    Page 16                                               GAO/RCED-91-26Drug Residuesin Milk
Federal Regulation of DNg ltmidues in the
=IJnPPlY




FM’S risk assessmentprocessfor animal drug residues in milk has six
steps:
1. Identifying the nature and amount of residues from the parent drug
and its metabolites and the depletion of the residues after treatment.
2. Comparing the metabolism of the drug in the animal speciesproposed
for the toxicity testing with the metabolism of the drug in the animal
targeted for treatment (to determine which residues must be tested for
toxicity and which laboratory animal speciesto use for toxicity studies).
3. Determining the toxic effects of residues,if any, from sponsor-sub-
mitted studies and the safe concentration of all residues from the pro-
posed drug.
4. Identifying which residue to measure as an indicator of the safe con-
centration of all residues and establishing a tolerance for that residue
(referred to as the “marker residue”).

6. Evaluating the analytical method proposed by the sponsor to measure
the marker residue reliably in milk at the tolerance level (referred to as
the “regulatory method”).
6. Establishing the withdrawal period for the administered drug to
ensure the depletion of residues at or below the safe concentration level
as indicated by the marker residue (when the marker residue is at or
below its tolerance level) so the milk can be safely consumed.(FDA does
not establish a tolerance when residues above the safe concentration are
unlikely to occur at zero withdrawal.)
For non-carcinogenicdrugs, FDAestablishesthe safe concentration for
the drug residues basedon the acceptabledaily intake (ADI) for the drug.
Adjustments are made for differences between test animals and humans
in food consumption versus body weight, a safety factor, and the esti-
mated amount of milk consumedby an individual per day. The ADI is the
estimated daily intake of a drug residue which, during a lifetime of
exposure (70 years), is not expected to causeappreciable health risks on
the basis of all known facts at the time. The ADI is basedon the highest
drug dosagedemonstrated to have no observableeffect in the most sen-
sitive test animal speciesused in the toxicity studies divided by a safety
factor.




Page 19                                     GAO/RCED91-28Drug Residuesin Milk
Appendix   I
Federal Ragulatlon of Drug Iteaiduea in the
-SuPPlY




The safety factor is intended to provide a margin of safety and account
for the inherent uncertainty in projecting the results of animal toxi-
cology tests to humans. FDA usessafety factors of 100 to 1,000
depending on the nature of the effects observed at the higher dosage
levels in the test animals and the length of the study.

FDAthen calculates the safe concentration using the ADI, the weight of
the averageadult (60 Kg), and the maximum amount of milk that FDA
estimates an individual can consumeper day-l .6 liters. FDA'S calcula-
tions assumethat all dairy cows are treated with the drug and thus an
individual will consume1.6 liters of milk containing the drug residues
per day, assumptions which FM officials believe overstate a person’s
likely exposure to actual drug residues in the milk supply.
For drug residues that are carcinogens,FDAdetermines the dosethat will
satisfy the “no residue” requirement of FFDCA   to establish the safe con-
centration level. Under FFDCA,FDAcannot approve a new animal drug if
the drug induces cancer in humans or animals unless (1) the drug will
not adversely affect the animals for which it is intended and (2) no res-
idue of the drug is found by methods approved by FDAregulation, in any
edible portion of the animals after slaughter or in any food derived from
the living animals. FDA has operationally defined “no residue” to mean
no significant risk of cancer- correspondingto a risk to test animals of
no more than 1 in 1 million over a lifetime of drug exposure. FIX calcu-
lates the concentration of drug residue yielding no significant risk of
cancer level from the tumor data using a statistical extrapolation
procedure.

Unlike the risk assessmentprocessused for other health effects of
drugs, FDA doesnot use an acceptabledaily intake in assessingcarcino-
genic effects. The ADIis a level of drug intake which is safe within a
practical certainty. Scientists have been unable to determine whether a
safe, threshold level exists for carcinogensbecausethe mechanismsthat
produce cancer are not completely understood. Therefore, lacking infor-
mation establishing the mechanismof carcinogenesisfor a particular
drug residue, FDAusesdose-responsemodels which assumethat some
risk of contracting cancer exists for even minute exposuresto carcino-
genic drug residues.Dose-responseassessmentdefines the relationship
between estimated dietary exposure to a carcinogenand the probability
of carcinogeniceffects.
There is no tolerance established for sulfamethazine (SMZ)in milk
becauseFDA has not approved its use in milk-producing dairy cows. If a


Page20                                        GAO/RCELk91-26Drug Residuesin Milk
                                                                                                          c

                          Federal   ItqulatIon   of Dmg Reaiduealn the
                          hfuk SUPPlY




                          veterinarian or farmer usesSMZin dairy cows, illegal drug residuesin
                          milk can result. Recently, ells estimated that treatment of just a single
                          cow with SMZ can contaminate the milk, when pooled, of 70,000 cows.


Concern Levels            When tolerances do not exist or cannot be calculated becausethe neces-
                          sary toxicology and residue data are not available, m may set “concern
                          levels” for drug residues.For milk, these levels correspondto the lowest
                          level of the drug residue that can be quantified by an analytical method
                          or one-third of the lowest published tolerance for the drug residue in
                          edible tissue.
                          Concern levels are not official tolerances and do not represent FDA
                          approval of the drug use. Rather, these values represent an informal
                          level of action/concern that FDAusesas a target for developing analyt-
                          ical methods to monitor unapproved usesand to help set priorities for
                          possible regulatory action against those who illegally use the drug.
                          In the past, FDAhas approved certain animal drugs and established a
                          zero tolerance for their residues,basedon the specified analytical
                          method available at that time. The sensitivity level of the available
                          method then, in essence,becamethe “action level” for these residues.
                          For example, m initially set a zero tolerance for erythromycin, an
                          antibiotic used to treat infections in dairy cows. However, the lowest
                          level quantifiable in milk using the analytical method acceptedwhen the
                          zero tolerance was established was 50 parts-per-billion. Consequently,
                          for regulatory purposes,F~DAhas “operationally” defined this tolerance
                          at 60 ppb.
                          m’s Center for Veterinary Medicine (CVM) has determined that safe
                          levels cannot be established for the drugs chloramphenicol and SMZ,
                          becauseof human food safety concerns.FI~Apolicy states that neither of
                          these drugs should be used in dairy cows becausethere is concern about
                          any detectable level of the drugs in milk. However, SMZ,which is avail-
                          able over-the-counter, is approved for cattle, dairy cows that are not
                          producing milk, swine, chickens, and turkeys in order to treat respira-
                          tory diseasesand, in someinstances,to promote weight gain.


Sulfamethazine Level of   SMZresidues in milk and other food products have becomea concernto
Concern U                 m becausethe drug has been shown to be carcinogenicin laboratory
                          animals. Two animal studies conducted by FDA'SNational Center for
                          Toxicological Researchhave demonstrated that moderate-to-high doses


                          Page 21                                        GAO/RCEl%91-26Drug Residuesin Milk
                         Appendix 1
                         FedetralRegulation of Drug lb&ha      in the
                         mk SUPPlr




                         of SMZproduced thyroid tumors in laboratory rats and mice. m is con-
                         sidering whether to permit continued use of SMZ as currently approved.
                         Controversy exists within m as to whether a safe level can be set for
                         SMZ becausethe data necessaryto make such a determination are incom-
                         plete. However, CVM has calculated a concern level for total residues of
                         SMZ in milk to be 12 ppb(for the parent compound and its metabolites
                         combined), on the basis of a preliminary risk assessmentusing available
                         toxicology data. CVM estimates that total SMZ residues of 12 ppbover a
                         lifetime of exposure may present no more than an insignificant risk of
                         cancer in humans. However, existing safety data do not allow a marker
                         residue to be established. CVM estimates that if the parent drug SMZwas
                         used as the marker residue, then the concern level would be in the 1 to 6
                         ppb range. CVM believes that the low levels of SMZ projected to be safe in
                         milk will preclude practical use of the drug in dairy cows.


Approved and             Residuesof drugs, including antibiotics, can occur in milk as the result
Unapproved Drug Use in   of legal or illegal use of drugs. FDAhas approved 63 drugs for use in or
                         on dairy cows, including 20 antibiotic drugs. FRAhas established toler-
Dairy Cows               ancesin milk for 21 of the 63 drugs approved for use in dairy cows3 In
                         addition, about 26 drugs, including 12 antibiotics, not approved for use
                         in dairy cows are believed to be used in the dairy industry. Generally, an
                         illegal use occurs when a drug residue is found that exceedsits tolerance
                         level, when misuse of a drug approved for use in dairy cows results in
                         residues in milk for which no tolerance has been established, or when
                         residues of a drug not approved for use in dairy cows results in residues
                         in milk.

                         Under FFDCA, the actual or intended use of a new animal drug in a food-
                         producing animal in a manner inconsistent with the approved labeling
                         causesthe drug to be adulterated. FDA may consider taking regulatory
                         action against the veterinarian, producer, or other persons involved,
                         whenever such actual or intended unapproved use is found.



                         3F’DAhas not establishedtolerancesfor many drugs approvedfor usein or on dairy cowsbecause
                         somedrugs were approvedyears agoon the basisof data that indicatedthat no safety problemwould
                         result from useof the drug, or becauseFDAdeterminedthat residuesof the drug aboveits safe
                         concentrationlevel would be unlikely to occur at zerowithdrawal. Also, tolerancesin milk exist for
                         four drugs,for which the approval of their usesin dairy cowshas apparently beenwithdrawn. FDA
                         attorneysadvisedus that failure to removethe tolerancesof thesefour drugs from the regulations
                         was inadvertent, and that FDAdoesnot establishtolerancesfor residuesin milk of drugsthat are not
                         approvedfor usein dairy cows.



                         Page 22                                                 GAO/lKXD-91-26 Drug Residueain MUk
    Appendix   I
    FederalBegulationofDrueBesiduesinthe
    =suPPlY




    FDA’sCenter for Veterinary Medicine has established guidelines for vet-
    erinarians to treat food-producing animals with drugs not approved for
    them, and/or not approved for the particular manner in which used, if
    the animal’s health is otherwise immediately threatened or suffering
    and/or death would result from not treating the affected animal. In
    establishing this policy, known as the extra-label drug use policy, FDA
    said that it would ordinarily refrain from initiating regulatory action
    against licensed veterinarians for using or prescribing in their practice
    any drugs they could legally obtain, provided
l a careful medical diagnosis was made by an attending veterinarian
  within the context of a valid veterinarian-client-patient relationship;
. a determination was made that: (a) there is no approved drug specifi-
  cally labeled to treat the condition diagnosedor (b) drug therapy at the
  dosagerecommendedby the labeling has been found clinically ineffec-
  tive in the animals treated;
. procedures are instituted to assurethat the identity of the treated ani-
  mals is carefully maintained; and
. the time period for drug withdrawal prior to marketing meat, milk, or
  eggsis significantly extended; steps are taken to assurethat the
  assignedtime frames are met; and no illegal residues occur.
    FDA’sextra-label use policy doesnot permit non-veterinarians, such as
    dairymen, to treat food-producing animals with drugs not approved for
    them and/or in an unapproved manner. FDA’spolicy states that lay per-
    sons cannot be expected to have sufficient knowledge and under-
    standing concerning animal diseases,pharmacology, toxicology, drug
    interactions, and other scientific considerations to use drugs in treating
    food-producing animals in any manner other than as labeled on an
    approved drug.
    FIN has declared that chloramphenicol and diethylstilbestrol may not be
    used in treating food-producing animals under the extra-label use policy.
    In addition, dimetridazole, ipronidazole, or other nitroimidazoles may
    not be used under the extra-label use policy in unapproved species.Fur-
    thermore, FDAhas required manufacturers of all SMZ products to add a
    warning to product labels that SMZ is not to be used in female dairy
    cattle 20 months of age or older.




    Page23                                     GAO/RCED-Vl-26DrugResiduesinMilk
                  Federal ltagulation of Drug R.&dues in the
                  Milk SUPPlY




                  FDAadministers the Federal/State Milk Sanitation Program through
Monitoring Milk   Interstate Milk Shippers Agreements to ensure the safety and whole-
Safety            somenessof fresh milk and cream in the United States. Under this pro-
                  gram, the producers of Grade A pasteurized milk are required to pass
                  inspections and be rated by cooperating state agencies.
                  FDA’smilk safety program is a collaborative federal/state effort that
                  dates back to the mid-1920s. The program was established after the pro-
                  mulgation of the Standard Milk Ordinance by the Public Health Service
                  to assist states and municipalities in initiating and maintaining effective
                  programs for the prevention of milk-borne diseases.
                  To provide for uniform interpretation of this ordinance, an accompa-
                  nying codewas published in 1927 which set forth administrative and
                  technical details to achieve satisfactory compliance. This milk regula-
                  tion, now titled the Grade A Pasteurized Milk Ordinance (the Milk Ordi-
                  nance), has undergone numerous revisions since that time and is the
                  basic standard used today in the voluntary cooperative interstate milk
                  safety program in which all 60 states and the District of Columbia par-
                  ticipate. The Milk Ordinance is recognizedby public health agencies,the
                  milk industry, and many others as a national standard for milk
                  sanitation.

                  Under the cooperative federal/state milk safety program, FDAdoesnot
                  routinely analyze milk samples for animal drug residues.Instead, FDA
                  relies on the states to do routine testing of the milk supply. Milk proces-
                  sors also routinely test raw milk. FDAdoesnot routinely review state or
                  processortest results, but has conducted sampling and testing for pesti-
                  cide chemicals,microbiological contaminants, and certain drug residues
                  basedon inspection findings or reports of potential problems.
                  Within the FDA,the administration of the agency’smilk safety program
                  is basically divided between the Center for Food Safety and Applied
                  Nutrition (cm) and CVM, with FDA’sfield offices performing inspec-
                  tions and sample collections and providing analytical support, CFSAN’S
                  Milk Safety Branch is responsible for monitoring the overall conduct of
                  the milk safety program carried out by the states. CVM is responsible for
                  providing technical expertise in the development of testing and analyt-
                  ical methodology related to animal drugs. In addition, CVM is responsible
                  for evaluating the safety and effectiveness of new animal drugs and,
                  with respect to new animal drugs for dairy cows, evaluating the condi-
                  tions for use that would preclude the presenceof potentially hazardous
                  residues in milk.


                  Page 24                                      GAO/WED-91-26Drug Residuesin Milk
Appe,ndix II

Resultsand Limitations of FDA’s Efforts to
Determinethe Presenceof Antibiotic Drug
Residuesin the Nation’s Milk Supply
                       FW conducted several efforts to determine the presenceof antibiotic
                       drug residues in the nation’s milk supply between 1988 and 1990,
                       including:
                   l In March 1988, FDAconducted a survey of 10 cities to determine the
                     presenceof sulfamethazine, a suspectedcarcinogen(cancer-causing
                     agent), in milk.
                   . Following the 1988 survey, FDA used data from a questionnaire sent to
                     state regulatory agenciesby the National Conferenceon Interstate Milk
                     Shipments to determine whether the presenceof SMZ in milk declined
                     between May 1988 and September 1988, following FDA'S efforts intended
                     to eliminate SMZ use in milk-producing dairy cows (dairy cows).
                   l In late 1989 and early 1990, FDAconducted a survey of 14 cities to deter-
                     mine the presenceof selectedantibiotic drug residues in the milk supply
                     after two independent surveys reported finding numerous contaminated
                     milk samples.

                       Although similar in purpose, each of FDA’Sefforts were different in
                       design and produced different results. In addition, limitations in FDA'S
                       efforts, both individually and collectively, may preclude any compari-
                       sonsof the results of the efforts and any conclusionsabout the safety of
                       the overall milk supply.


1988 Survey            researchersdetected the residues of a variety of animal drugs in milk.
                       Among the drug residues reportedly detected was SMZ, one of about 46
                       antibiotic drugs in the class of drugs known as sulfonamides (sulfa). SMZ
                       use in animals has been controversial becauseit is a suspectedcarcin-
                       ogen. In addition, SMZ residues in milk may pose a risk for individuals
                       allergic to sulfa-based drugs.
                       Although SMZ was not approved for use in milk-producing dairy cows
                       and FDAhad not established tolerances for SMZ residues in milk, FDA
                       established an unofficial concern level for SMZ at 10 ppb in milk. Subse-
                       quent to reports that independent surveys had detected SMZ in milk, FDA,
                       in March 1988, conducted its own milk survey related to the presenceof
                       SMZ.



Methodology    *       FDAregional office personnel collected five retail shelf milk samples
                       (representing five different dairy processors)from each of 10 cities



                       Page 26                                   GAO/RCED-91-26Drug Residuesin Mtlk
                 AppendixII
                 Redta and Llnd~tions of FM% Effort43to
                 Detennlne the Reeence of Antibiotic Drug
                 Residueain the Nation’s Milk Supply




                 (Atlanta, Boston, Chicago,Dallas, Denver, KansasCity, Newark, Phila-
                 delphia, San Francisco, and Seattle). FDAscientists initially used a test
                 method, known as high pressure liquid chromatography (HPLC), to
                 screena total of 49 samples(one location provided only four samples)
                 for SMZ.FDAscientists used a second,more specific method, known as
                 mass spectrometry, to confirm the presenceof SMZin the samplesthat
                 yielded positive results over 10 ppb under the HPLCmethod.


Results          The March 1988 survey found varying levels of SMZin 73 percent of the
                 samplestested. Specifically, 36 of the 49 samplestested showed levels
                 of SMZpresent in the milk ranging from 0.8 ppb to 40.3 ppb. Five samples
                 tested above 10 ppb,the unofficial concern level for SMZthat FDAwas
                 ‘using at that time, and 10 of the samplestested above 6 ppb.


Limitations      The 1988 FDAsurvey was limited for several reasons.First, the survey
                 focused only on one animal drug-SMZ. The survey did not determine
                 whether any other unapproved drug residues were present in the milk
                 samples although FDAofficials believe that about 26 unapproved drugs,
                 someof which may posetoxicological concernsto humans, might be
                 used in the dairy industry. The survey also did not determine whether
                 any approved drugs were present at levels above their tolerances.
                 Second,FDAdid not design the survey to provide any statistically valid
                 estimates from the survey results to the nation’s milk supply. Thus, no
                 conclusionscan be reached on the basis of this limited survey regarding
                 the safety of the nation’s milk supply.

                 The March 1988 survey results raised FDA’Sconcern about the possible
1988 Follow-Up   misuse of SMZby the dairy industry becauseSMZwas not approved for
Survey           use in dairy cows. In response,J?DAtook several steps to eliminate SMz
                 residues in the milk supply, including an educational campaign aimed at
                 dairy farmers. FDAcoordinated this educational effort with the National
                 Conferenceon Interstate Milk Shipments (the National Conference)-a
                 voluntary organization comprised of federal and state health and agri-
                 cultural officials and the dairy industry-that together with FDA,over-
                 seesa cooperative, federal/state program (the Interstate Milk Shippers
                 Program) to ensure the sanitary quality of milk and milk products
                 shipped interstate.
                 To assessthe effectiveness of the educational program and follow-up on
                 the 1988 FDAsurvey, the National Conferencesent a questionnaire on


                 Page 26                                    GAO/RCED-91-26Drug Residuesin Milk
              Resulti and undtations of FM% Rfforta to
              De&erndnethe Rwence ofAntibiotlcDrng
              Residuesin the Natlon’r MUJcSupply




              October 8,1988, to all state regulatory agencylaboratories to obtain
              information regarding the number of raw milk samplestested for SMZ
              from May 1988 to September 1988 and the results of those tests. FDA’S
              Milk Safety Branch analyzed the results.


Results       FDA’sanalysis of state data showed that 4,887 samplesof raw milk were
              tested from May to September 1988 and that 6 percent of the reported
              samplestested (247 out of 4,887) contained SMZ.Further, FDAreported
              that only 1 percent of the reported test samples(64 out of 4,887) con-
              tained SMZresidues above 10 ppb.According to FDA,the results of the
              follow-up questionnaire indicated a significant reduction in the presence
              of SMZ residues in milk. On the basis of these results, FDAdeclared that
              the SMZ problem had been resolved.


Limitations   Although FDAconcludedthat the results from the follow-up survey
              showed a dramatic decreasein the level of SMZresidues in the nation’s
              milk supply compared with the 1988 FDAsurvey results, limitations in
              the follow-up survey preclude direct comparison with FDA’S1988 survey
              or any conclusionsregarding the safety of the nation’s milk supply.
              The follow-up survey cannot be relied upon to be representative of the
              nation’s milk supply. First, the states responding to the survey who
              tested milk samplesproduce only about 66 percent of the nation’s milk
              supply. More importantly, FDAcould not provide documentation showing
              how the responding states sampled milk products. If the testing states
              did not use statistical methods to select the samples,then the resulting
              state and national data are highly suspect.FDAalso did not know
              whether similar analytical methods and calibration standards were
              employed by the states in testing for SMZresidues,or whether all the
              methods used were capable of detecting SMZat the 10 ppbhealth concern
              level FDAhad established at that time.
              In particular, somestates may have used a method that is unable to
              detect many drug residues, including SMZ,at their levels of concern.
              Under the Pasteurized Milk Ordinance,the Bacillus Stearothermophilus
              disk assay is the only official method recognizedto detect antibiotic
              drug residues in milk for regulatory purposes.However, FDAscientists
              have found that this method is primarily useful only in detecting peni-
              cillin antibiotics and cannot detect low levels of many other antibiotics.
              For example, the method can only detect levels of SMZat and above 16



              Page 27                                    GAO/RCED-Sl-26DcuglteeidueninMllk
             Appendix II
             Result8 and Lllnltations of FM’s lwforta to
             Detennlne the Presenceof Antibiotic Drug
             Residuesin the Nation's Milk Supply




             parts-per-million (ppm) or higher- 1,600 times the 10 ppb concern level
             set by FM for SMZin milk.
             Although FDAofficials believe that states had other testing methods
             available in addition to the disk assay method to detect SMZresidues, it is
             unknown what specific methods the states used. If somestates used
             only the disk assay method, then their results might have understated
             the number of samplescontaining SMZresidues above FDA’Slevel of con-
             cern. Also, similar to the earlier 1988 survey, the 1988 follow-up effort
             only obtained data on the presenceof sMz-no data on the possible pres-
             enceof other unapproved or approved drugs was gathered.
             We cannot say with certainty that the results developed by FDAin its
             follow-up survey are flawed. However, given a sample representing only
             66 percent of the nation’s milk supply, no assurancesthat state samples
             were statistically drawn, and questions about state testing compara-
             bility, the results would likely be correct only by coincidence.

             Subsequentto FDAconcluding, on the basis of the 1988 follow-up survey,
1990Survey   that the SMZproblem had been solved, the Wall Street Journal (the
             Journal) reported the results of two surveys of animal drug residues in
             milk on December29,1989. One survey was sponsoredby the Journal
             and the other by the Center for Sciencein the Public Interest (CSPI), a
             consumer food safety and nutrition organization.

             The Journal reported that its survey found that 38 percent of 60 retail
             milk samples contained residues of antibiotics, possibly including SMZ
             and other unapproved drugs. The CSPIsurvey found that 20 percent of
             20 retail milk samplescollected in the Washington, DC., area contained
             sulfa drugs, again possibly including SMZand other unapproved drugs.
             Both surveys used an analytical method called Charm II.
             The Charm II test is considered a screeningtest becauseit can report-
             edly detect the presenceof sevenclassesof antibiotic drug residues in
             milk (e.g., aminoglycosides,beta-lactams (penicillins), chloramphenicol,
             macrolides, novobiocin, sulfonamides, and tetracyclines). However,
             except for chloramphenicol and novobiocin, which are individual chem-
             ical entities, the Charm II test can only indicate that a member of a par-
             ticular chemical family of antibiotics may be present in milk samples
             tested, it cannot identify the specific antibiotic drug residue(s) respon-
             sible for the positive result. The identification of the specific antibiotic
             drug(s) must be determined by an independent confirmatory method of


             Page 28                                       GAO/RCEDBl-26 Drug Residuesiu Milk
              Appendix   ll
              Remlta and Limitationa of FDA’s Efforta to
              Deterndne the Presenceof Antibiotic Drug
              Residuesin the Nation’s Milk Supply




              analysis, according to FDAofficials. For example, the Charm II test can
              indicate that a member of the sulfonamide classof antibiotics may be
              present in a milk sample tested, but cannot identify which specific sul-
              fonamide drug(s), such as sulfamethazine or sulfadimethoxine, caused
              the positive response.


Methodology   rm~designedits 1990 survey to test the reliability of the independent
              surveys and to confirm or discount claims that animal drug residues are
              present in milk. To do this, FDAfield office personnel obtained two con-
              tainers of milk with the samelot number and date from 6 stores in each
              of 14 cities. The 14 cities included all those in the Journal’s survey, as
              well as those included in FDA’sprevious survey efforts (Atlanta, Balti-
              more, Boston, Chicago,Dallas, Denver, KansasCity, Los Angeles, Miami,
              Minneapolis, New York, Philadelphia, San Francisco, and Seattle). The
              milk samplescollected were then analyzed for the presenceof antibiotic
              drug residues.According to FDAofficials, the survey was a “snapshot”
              of the presenceof certain drug residues in milk.
              The containers collected at each location were shipped over night; one            .
              container was sent to FDA'SBeltsville laboratory and one container was
              sent to FDA'SPhiladelphia District laboratory. Beltsville and Philadelphia
              ultimately received 70 milk sampleseach-l container each from the
              five stores selectedin each of the 14 cities. The Philadelphia laboratory
              used the sameCharm II test method used in the Journal and CSPIsurveys
              to screenits 70 milk samplesfor drug residues.Subsamplesof all sam-
              ples found positive by the Charm II test in Philadelphia were sent to
              FDA'SDenver laboratory for further testing using modified HPLC,thin
              layer chromatography, gas chromatography, and microbiological test
              methodologies.Denver also performed mass spectrometry confirmatory
              testing for samples showing positive screeningresults for chloramphen-
              icol, sulfadiazine, and sulfamethazine.

              FDAofficials believe that the mass spectrometry method, which is more
              difficult and costly to perform than other methods, represents state-of-
              the-art procedures in analytical chemistry and is the most reliable
              method available for identifying specific compounds.FDA considersall
              other test methods, including HPLC,to be screeningtests because
              although they may be able to tentatively identify which one of a number
              of drugs may be present in milk, they cannot positively identify the drug
              found.




              Page 29                                      GAO/RCED-91-26Drug Residuesin Milk
                        ltaaulta and rdndtationa of FlM% Effort49to
                        De&ma&e the Presenceof Antibiotic Drug
                        Residuesin the Nation’e MU Supply




                        The Beltsville laboratory used multi-residue HPW screeningmethods to
                        test all of its 70 samples for the presenceof 10 sulfa and 3 tetracycline
                        antibiotic drugs. The Beltsville laboratory also used mass spectrometry
                        to confirm and “fingerprint” the results of their HPW testing, but only
                        had confirmatory tests for chlortetracycline, oxytetracycline, sulfadia-
                        zine, sulfamethazine, and tetracycline. Also, with the exception of three
                        sulfamethazine samples,Beltsville limited its confirmatory testing to
                        samplesthat its HPLCtesting indicated had residues equal to or in excess
                        of established concern levels.
                        Overall, FDAcould only perform confirmatory testing for six of the
                        antibiotics selectedfor the 1990 survey, and, according to the Director
                        of FIN’SCenter for Veterinary Medicine, the survey cost $360,000.


Results                 In total, the screeningtest results from FDA’sthree participating labora-
                        tories initially indicated that 91 percent of the milk samplestested (64
                        out of 70) contained low levels of antibiotic drug residues and that many
                        samplescontained multiple drug residues.FDAperformed mass spec-
                        trometry testing on three of the samplesfor which screeningtests indi-
                        cated low levels of sulfamethazine and confirmed its presencebelow 6
                        ppb. FLWdid not find any antibiotic drug residues at or above concern
                        levels, or at or above the level of detection sensitivity of the methods
                        used on the remaining samplesselectedfor mass spectrometry testing.
                        However, FDA’sattempt to confirm the presenceof
                        sulfachloropyridazine was inconclusive becauseof problems with the
                        confirmatory method. FDAonly had confirmatory methods for six of the
                        other drugs selectedfor the 1990 survey, and three of these methods
                        could not detect their respective drugs at established concern levels.


Philadelphia/Charm II   The Philadelphia laboratory’s results using the Charm II test were sim-
Results                 ilar to the results from the Journal and CSPIsurveys in that many of the
                        samplestested indicated the presenceof antibiotic drugs. The Philadel-
                        phia laboratory results indicated that 61 percent (36 out of 70) of the
                        samplescontained antibiotic drug residues,mostly tetracyclines, and
                        somesamplescontained multiple residues. Table II. 1 provides a sum-
                        mary of FDACharm II results.




                        Page 30                                       GAO/IUXD-91-26 Drug Residuesin Milk
           .


                                        AppendlxII
                                        RemlltaandLlml~tlona of FM% Efrortato
                                        Ik3tellulnethe-        ofAnubloticDrug
                                        ResldaeaIn the Nation’sMilk Supply




Table 11.1:FM Screonlng Rerulto for
Drug Frmlly Rorlduer In the 1990 Milk                                                                         Percent of
Suwoy Uolng the Charm II Tw                                                               Number of   Positive Safnosll;
                                                                                  Poeltlve SaFot;;
                                        Drug Family                                                          (out of 70)
                                        Aminoglycosides (gentamicin)                              4               --- 6
                                        Bets-lactams (penicillin)                                 1                   1
                                        Chloramohenicol                                           1                   1
                                        Macrolides (erythromycin)                                 1                   1
                                        Novobiocin                                                5               --  7
                                        Sulfonamides                                             15                  21
                                        Tetracvclines                                            34                     49
                                        Source: Prepared by GAO from FDA data.

DenverLaboratory Results                Subsamplesof the 36 samples found positive using the Charm II test
                                        were sent to m’s Denver laboratory for further screeningand confir-
                                        matory testing. Basedon this screeningand/or confirmatory testing, the
                                        Denver laboratory reported that it could not corroborate any of the Phil-
                                        adelphia laboratory’s positive Charm II results. However, the test
                                        method Denver used for five samples which Charm II indicated the pres-
                                        enceof novobiocin was not as sensitive to the drug as the Charm II test
                                        is claimed to be. The Denver method was also incapable of detecting
                                        novobiocin residues at the health concern level established by FDA.
Ekltsville Laboratory Sulfa Drug        The Beltsville laboratory’s HPU=test results indicated that 86 percent
Resulti                                 (60 out of 70) of the milk samplestested contained sulfa drug residues;
                                        46 percent (32 out 70) of the samplescontained multiple sulfa residues.
                                        Specifically, Beltsville initially found that 16 percent (11 out of 70) con-
                                        tained sulfa residues greater than or equal to 10 ppb, the FDA-designated
                                        concern level for these drugs at that time; 83 percent (68 out of 70) con-
                                        tained sulfa residues at levels less than 6 ppb;and 1 percent (1 out of 70)
                                        contained residues between 6 and 10 ppb. Table 11.2provides a summary
                                        of initial results from the Beltsville laboratory using HPLC.




                                        Page31                                     GAO/RCEDBl-26
                                                                                               DrugResiduesin Milk
                                            Resulta and Limitation of FDA’s Eff-ortu to
                                            Detmudne the Presence of Antibiotic Jhug
                                            IUeiduen in the Nation% Milk Supply




Table 11.2:lnltlal FDA Screening Retrultr
for Sulfa Drug Realdues in the 1990 Milk                                                                              Percent of
Survey Ming HPLC                                                                                  Number of   Poritlve Sapot;;
                                                                                          Poaitlve Saro$l;
                                            Drug                                                                     (out of 701
                                            Sulfachloropyridazine                                         3                   4
                                            Sulfadiazine                                                 32                  46
                                            Sulfadimethoxine                                              3                   4
                                            Sulfamerazine                                                47                  67
                                            Sulfamethazine                                               10                  14
                                            Sulfamethizole                                                1                   1
                                            Source: Prepared by GAO from FDA data.

                                            Beltsville’s HPU results also indicated that 10 of the 11 samplesthat
                                            tested positive for sulfa residues at or above 10 ppbcontained sulfadia-
                                            zine at levels ranging from 10 to 316 ppband that 1 sample contained
                                            sulfachloropyridazine at 16 ppb.
                                            However, the Beltsville laboratory did not confirm the presenceof sulfa-
                                            diazine at or above 10 ppb in any of the samples found positive by their
                                            HPLCmethod. Through additional testing using a revised HPU procedure,
                                            thin layer chromatography, and mass spectrometry, Beltsville subse-
                                            quently found that the compound theobromine interfered with the ini-
                                            tial HPLCscreeningtests and led to false positive results. Theobromine is
                                            a caffeine-related substancefound in chocolate.FDAbelieves that the
                                            substancemay have found its way into the milk sampled as a trace res-
                                            idue of chocolate milk processedby the samedairy plants. (Beltsville
                                            did not reanalyze sampleswith low levels of sulfadiazine (less than 10
                                            ppb) using the revised HPLCprocedure becausethese levels were under
                                            the designated level of concern.) Beltsville attempted to confirm the
                                            presenceof sulfachloropyridazine above 10 ppb in the remaining sample,
                                            but its analysis was inconclusive becauseof problems with the confir-
                                            matory method.
                                            Although low levels of sulfa drugs (i.e., less than 5 ppb)were detected in
                                            many of the milk samplesusing the HPLCmethod, FDA advised that these
                                            results must be interpreted cautiously becausemany low level positive
                                            results could be due to interference near the ability of this method to
                                            detect low levels of sulfa drugs in milk. Also, Beltsville generally did not
                                            attempt to confirm HPLCresults indicating sulfa residues below the con-
                                            cern level of 10 ppb.However, three sampleswith the highest indicated
                                            concentrations of SMZ(all less than 6 ppb) were analyzed using the mass



                                            Page 32                                        GAO/RCED-91-26Drug Residuesin Milk
                                Appendix II
                                Reeulta and Lhitatlone of FM’II Rfforta to
                                Detmntne the Prewnce of Antibiotic bye
                                Reelduesin the Nation’s Milk Supply




                                spectrometry method and the presenceof SMZ~in these sampleswas
                                confirmed.

Beltmille Laboratory            The Beltsville laboratory HPLCtest results indicated that one sample
Tetracycline Results            tested contained oxytetracycline at 99 ppb.Beltsville was unable to con-
                                firm the presenceof this drug using its mass spectrometry method.
                                However, the confirmatory method could only detect oxytetracycline at
                                100 ppb or more. The health concern level FDAset for this drug was 30
                                ppb.


Limitations                     As in its previous efforts, the 1990 FDAsurvey of antibiotic drug resi-
                                dues in milk was not statistically designedto be representative of the
                                nation’s milk supply. According to rn~ officials, the survey was a snap-
                                shot and the results could be completely different if the samesurvey
                                was conducted again. In addition, the 1990 survey was limited for sev-
                                eral reasons.

Limit.& Number of Confiiatiry   FDAmay have been unable to confirm someof the positive screeningtest
MethodsAvailable                results in its 1990 survey becauseit did not have methods to identify
                                and confirm the presenceof all drugs detected by the screening
                                methods. Thus, unconfirmed results may have been due, in part, to the
                                presenceof a drug residue not detectable by FDA’S confirmatory method.

                                For example, about 32 antibiotic drugs, approved and/or unapproved
                                for use in dairy cows, may have been in use by the dairy industry at the
                                time of the 1990 survey. However, other than the Charm II test, FDAonly
                                had screeningtests of its own for 17 antibiotic drugs included in its 1990
                                survey and mass spectrometry methods to confirm the presenceof only
                                six of these drugs in the milk samplestested. Regarding sulfa antibi-
                                otics, the Charm II test FDA’s Philadelphia laboratory used reportedly
                                could, in its screeningmode, detect 15 drugs in the sulfa class family.
                                Only 10 sulfa drugs were included in the 1990 survey, and FDAhad con-
                                firmatory (mass spectrometry) methods for only 2 of these.
                                In addition, the Beltsville laboratory found evidenceof a substancepre-
                                sent in milk samplescollected from Philadelphia that did not correspond
                                to any of the 10 sulfa drugs that its HPLC methods could identify.
                                According to the Chief, Method Validations and Analytical Branch,
                                ‘During the 1990survey, CVM revisedits preliminary risk assessment   for SMZand concludedthat it
                                could not set a safelevel of SMZin milk becauseof data gaps,but estimateda safeconcentrationof
                                total residuesof SMZ(including metabolites)in milk to be 12 ppb, with a likely markerresiduein the
                                l-6 ppb range.



                                Page 33                                                  GAO/RCED-91-26Drug Residuesin Milk
                                                                                                      .
                               Redta and Lbdt.atione of me Effort# to
                               DetermIne the Rwence of Antibiotic Drug
                               Reaidnw in the Nation’6 Milk Supply




                               Center for Veterinary Medicine, the presenceof the identified substance
                               could possibly be due to the processingand packaging of the retail milk
                               samplesused in the survey, the result of animal husbandry practices, or
                               the presenceof another sulfa or someother drug.

DiscountedResults Questioned   FW discounted Charm II test results from its Philadelphia laboratory
                               that indicated the presenceof novobiocin in Ssamples, or 7 percent of
                               the 70 samplestested, basedon the results of a microbiological screening
                               test method used by its Denver laboratory that was less sensitive to the
                               drug than the reported detection capability of Charm II. The microbio-
                               logical method was also unable to detect novobiocin residues at the tol-
                               erance level established by rn~.

                               Specifically, using the Charm II test with a reported detection sensitivity
                               of 60 ppb,the Philadelphia laboratory found five samplesthat contained
                               novobiocin. Denver, using a “green book” microbiological screeningtest
                               with a detection sensitivity of 200 ppb,did not corroborate these results.
                               However, the tolerance level established by FQAfor novobiocin was 100
                               ppb. According to FDAofficials, there were no other screeningtests avail-
                               able for novobiocin. Also, according to the Chief, Method Validations
                               and Analytical Branch, Center for Veterinary Medicine, the Beltsville
                               laboratory did not test milk samples for this drug becauseFW lacked a
                               confirmatory method for novobiocin. Consequently, although the Charm
                               II results indicating the presenceof novobiocin were discounted, FIN did
                               not know whether novobiocin was present in the samplestested above
                               the established tolerance level, but below the detection level of the green
                               book method used.

Limitations in Existing        Someof the other analytical methods FDAused in the 1990 survey were
confirmatory Methods           also unable to detect and confirm the presenceor absenceof various
                               drug residues at their concern levels. For example, three of the six drugs
                               in the 1990 survey for which FW had confirmatory methods were tetra-
                               cycline antibiotics. Thesethree drugs were/are approved to treat dairy
                               cows, but only one, chlortetracycline, had an m-established tolerance
                               for its residue in milk. For survey purposes,F+DA established unofficial
                               concern levels for the residues of these three drugs in milk. However,
                               the confirmatory methods FDAused in the 1990 survey were incapable
                               of detecting the presenceof these drugs at their concern levels. For
                               example, the concern level for oxytetracycline was 30 ppb, but FRA’S
                               mass spectrometry confirmatory test could only confirm it at 100 ppb or
                               more. Consequently, although the Beltsville laboratory found one
                               sample that contained oxytetracycline at 99 ppb using the HPU=method,



                               Page 34                                    GAO/ICED-@l-36Drug Reoidueain Milk
Beltsville was unable to confirm the HPLCresult with the mass spectrom-
etry method.
Thus, although m reported that it could not confirm the presenceof
any tetracycline drugs, the drugs may still have been present at levels
exceedingthe concern level for human consumption, but below the con-
firmatory test level of detection for these drugs. Table II.3 shows a com-
parison between the tolerance/concern levels for the antibiotic drug
residues in milk tested and the level of detection of testing methods m
used in the 1990 survey.




Page36



                        ‘,,
                                                 Appendix II
                                                 Reeulte ad Llmltations of FM’8 Rfforta to
                                                 Detmmlue the Presenceof Antibiotic Drug
                                                 Reeidueain the NatIon’s Milk Supply




Table 11.3:Tolerances/Levels of Concern for Selected Antibiotic Drug Residues in Milk and the Level of Detection for the Methods
FDA Used In Its 1990 Survey (In Parts-Per-Billion)
                                           Tolerance/ Philadelphia         Beltsville                        Denver
Drug                                  Concern Level’     Charm Ilb      BHPLCC BMSd            DQBMW DHPLC’ DTLV            DMSh
Penicillin                                          o/10                4.8’                 .             .                10           l            l         l


Tetracycline                                       -/80                200’                50        100              50-175           40             9         l


Chlortetracycline
          __
                                                    o/30               200’                30        100              50-175           40             l         l


Oxytetracycline                                    -130                200’                40        100              50-l 75          40             l         l


Chloramphenicol                                    -IO                 100                  .          .                     .           .            .        1
  .._._           _..-  _.-._-...-.-__--__
Streptomvcin                                        O/125              1OOi                 .          .                     .           .            .        .
Gentamicin                                         -130                  50’                .          .               20-70             l            l         l


Erythromycin                                        o/50               100’                  .         .                   25            l            .         l


Novobiocin                                       1oo/-                   50                 .          .                  200            l            l         l
               ..~.._

Sulfanilamide
      _. “~“.        .._......_ --...  --____~
                                                   -110                  10’                 5             l
                                                                                                                                 .       .            .         .


Sulfadiazine                                       -/IO                  10’               0.9         10                        .       .      100            10
Sulfathiazole
          .   ..-.--.-..--
                                                   -/IO                  10’                 1             l                     .       .       50             ’


Sulfamerazine          ._. i--_-.     ._.----
                                                   -/IO                  IO’               0.5             l
                                                                                                                                 .       .       10             l


Sulfapyridine                                      -/IO                  10’               0.9             l                     .       .            .         .


Sulfamethizole
_ .._ _
                                                   -110                  IO’                 2             l                     .       .            .         .


Sulfamethazine                                     -110                  IO’                 2             2                     .       .            5        5
Sulfachloropyridazine                              -110                  IO’                 1             l                     .       .            .         .


Sulfadimethoxine                                  lO/lO                  10’               0.7             l                     .       .
                                                                                                                                                      5         l


Sulfaquinoxaline                                   -/IO                  10’                 1             l                     .       .
                                                                                                                                                 IO             9



                                                 Qalue indicates tolerance and/or level of concern for drug residues in milk at the time of the FDA 1990
                                                 survey.

                                                 bCharm II = Values are for the screening mode of the Charm II test used by FDA’s Philadelphia District
                                                 Laboratory.
                                                 %HPLC = High pressure liquid chromatography        method that the FDA Beltsville Laboratory used

                                                 %MS = The mass spectrometry       method that the Beltsville Laboratory used to confirm test results.

                                                 ‘DGBMB = FDA Antibiotic Residues in Milk, Dairy Products and Animal Tissues: Methods, Reports, and
                                                 Protocols, Revised Oct. 1968. Reprinted Dec. 1974. (Called the Green Book microbiology methods.)
                                                 ‘DHPLC = Modified HPLC method that the FDA Denver Laboratory used.

                                                 eDTLC = The thin layer chromatography      method that the Denver laboratory used.

                                                 hDMS = The gas chromatography/mass         spectrometry       method that the Denver Laboratory used.

                                                 ‘The Charm II test generally detects the presence of antibiotic drug families-not     individual drugs. The
                                                 values given represent the levels of detection for the family of drugs for the individual drug listed.

                                                 ‘Different Charm II detection levels are given for gentamicin and streptomycin although both are mem-
                                                 bers of the aminoglycoside family, because the manufacturer claims that the test is more sensitive to
                                                 gentamicin.
                                                 Source: Prepared by GAO using FDA data.




                                                 Page 36                                                            GAO/RCED91-26Drug Residueein Milk
                              lballt4 and Llmlmlone of ma J?iffort$to
                              Determine the Preeenceof Autibiotk Drug
                              Reeidur, in the Nation% Milk Supply




                              In establishing concern levels for the 1990 survey, FDAcould only con-
                              sider parent drug compoundsbecausethey lacked data on the metabo-
                              lites of the parent drug compoundsin milk. Accordingly, FDAdid not
                              analyze the milk samplesfor metabolites of the antibiotic drugs they
                              tested for-only the parent drug compound.

Unexplained Discrepanciesin   Several unexplained discrepanciesin test results exist from the different
ResultsExist                  methods that m used in its 1990 survey that may render the survey
                              results inconclusive. For example, as indicated earlier in table II. 1, most
                              of the Charm II positive results were for tetracycline. However, the
                              Denver laboratory was unable to corroborate these results using green
                              book microbiological methods, even though these methods were report-
                              edly more sensitive to tetracycline than Charm II. According to FDA,
                              there are no data available to explain this difference. Similarly, the Belt-
                              sville laboratory was unable to detect three specific tetracyclines in
                              most of the samplesthat Charm II found positive for tetracycline drugs,
                              even though the HPL~method Beltsville used was reportedly more sensi-
                              tive to these drugs than Charm II.

Validity of Test Methods      Questionsexist about whether the methods FDAused in its 1990 survey
                              had been adequately validated. The HPLCand mass spectrometry
                              methods FDAused in the 1990 survey did not undergo multilaboratory
                              evaluation as specified in the Center for Veterinary Medicine formal
                              methods trial procedures.For example, the Center considersthe HPLC
                              method only a research method becauseit has not gone through the
                              usual validation procedures normally followed in FDA.FDAofficials said
                              that the methods were validated at FDA'Sown laboratories and peer
                              reviewed by two university scientists expert in toxicological analysis.
                              Although the reviewers concludedthat the manner in which FDAused its
                              methods in the survey was credible and adequate,they could not fully
                              “credentialize” the procedures becauseof the lack of data on the repro-
                              ducibility of the methods. Furthermore, the scientists questioned the
                              number of false positives produced by the Beltsville HPLCscreening
                              procedures.
                              In addition, two scientists from the Centers for DiseaseControl (CDC)
                              evaluated FDA'Sanalytical methods used in the 1990 survey and found
                              them adequate. However, the cut scientists suggestedthat additional
                              method development work is neededto determine the performance of
                              the methods using retail versus raw milk and to lower the level of detec-
                              tion for the methods below the concern levels for the tetracycline drugs
                              tested.



                              Page 37                                     GAO/RCED-91-26Drug Residuesin Milk
                                                                                             *
                                                                                                  --
                           APpsndk   n
                           Reeulw and Llmt~tlonm of FMb Efforta to
                           DetonnlnetlKlPreeeneeofAnnbiotleDrug
                           Reeidue#lntlIe~wr~supply




                           For example, the analytical methods FDAused to detect the presenceof
                           the three tetracyclines were developedusing raw milk. However, FDA
                           collected milk samplesoff the shelf. The extent to which retail versus
                           raw milk may affect test results is unknown.

Handling of Milk Samples   Questionsexist about FW’Squality assuranceof the handling of the milk
c!dkted                    samplescollected. About 46 percent of the milk samplesanalyzed in the
                           1990 survey were beyond their shelf-life “pull-by dates.” According to
                           FW, milk conditions such as spoiled milk, can causethe Charm II test to
                           indicate false positive results, especially for tetracyclines. Consequently,
                           the selection and handling of milk samples are critical to achieving accu-
                           rate results.
                           According to FDAofficials, becauseof time constraints the 1990 survey
                           was not structured and no criteria were set for what the pull-by dates
                           for milk samples should have been. However, none of the samples ana-
                           lyzed by FDAwere spoiled, according to the Chief, Method Validations
                           and Analytical Branch, Center for Veterinary Medicine.




                           Page 80                                    GAO/I&ED-@l-26Lhqij Residuesin Milk
 *a
Appendix iI

ObJ-          'WS,   Scope,and Methodobgy


                         As requested by the Chairman, Human Resourcesand Intergovern-
                         mental Relations Subcommittee,HouseCommittee on Government Oper-
                         ations, we reviewed the adequacy of survey efforts conducted by FDAin
                         1988 and 1990 to determine the presenceof animal drug residues in milk
                         and whether the information developedprovided sufficient basis for
                         m’s public statements attesting to the safety of the milk supply.
                         To obtain information on FDA’ssurvey efforts, we interviewed officials
                         and obtained documents from FDA’sCenter for Veterinary Medicine,
                         Center for Food Safety and Applied Nutrition, Office of Regulatory
                         Affairs, and the FDAlaboratories at Heltsville, Denver, and Philadelphia.
                         We also reviewed documents the Subcommitteeobtained from FDA
                         related to its surveys of animal drug residues in milk and pertinent
                         reports FDAissued as a result of its surveys. In addition, we reviewed
                         FDA,state, and industry testimony regarding their efforts to detect drug
                         residues in milk, given before the Subcommitteeon February 6, 1990.
                         We also met with the Center for Sciencein the Public Interest, to deter-
                         mine its views of FDA'Ssurveys of animal drug residues in milk.

                         Our review, which was done from April to September 1990, was con-
                         ducted in accordancewith generally acceptedgovernment auditing stan-
                         dards. We conducted our review primarily at FDA'SCVMheadquarters in
                         Rockville, Maryland.

                         The information in this report was discussedwith officials in FDA'S
                         Center for Food Safety and Applied Nutrition, Center for Veterinary
                         Medicine, and Office of Regulatory Affairs. Where appropriate, changes
                         have been made basedon the discussionto further clarify the informa-
                         tion presented. However, as your office requested,we did not obtain
                         official agency comments on a draft of this report.




                         Page 39                                   GAO/RCED91-26Drug Residuesin Milk
Appendix IV

Major Contributors to This Report


                          Edward M. Zadjura, Assistant Director
Resources,                Michael J. Rahl, Assignment Manager
Community, and            Max L. Aguilar, Evaluator-in-Charge
Economic                  William M. Layden, Senior Evaluator
                          Scott W. Weaver,Staff Evaluator
Development   Division,                  - Statistician
                          Mitchell B. Kamman.
Washington, DC.




(160704)                  Page 40                                 GAO/BCED-91-28m   Residuea in Milk
Ordering   Ittformatiott

‘I’ttv first. five c0pir-s of each GAO report. are free. Additiottal copies
are $2 each. Orders should be sent to the followittg address, accom-
ptttitvl by a. check or money order made out. to the Superitttendettt
of I)ocutttt~tit.s, when necessary. Orders for 100 or ttlort* copies to be
tttailt~d LO a sittglv address are discounted 25 percent,.

Ii.% (;t~titval Accoutitittg Off&
P.O. 130x 6016
GaiUtc~rstmrg, MI) 20877

()rdc~rs may also be placed by calling   (202) 275-624 1.
r
    .-_.._ .-...   .-.-   -_.-   .___-   -_-... “--...“.._   ._._.-.-_..--...l   __.__-_   --.-_---__---   -.-.   -----   -_.