Embed Size (px)
Citation preview
Risk Analysis, Vol. 30, No. 3, 2010 DOI: 10.1111/j.1539-6924.2009.01340.x
Assessing Potential Human Health Hazards and Benefitsfrom Subtherapeutic Antibiotics in the United States:Tetracyclines as a Case Study
Louis Anthony (Tony) Cox Jr.∗ and Douglas A. Popken
Many scientists, activists, regulators, and politicians have expressed urgent concern that us-ing antibiotics in food animals selects for resistant strains of bacteria that harm human healthand bring nearer a “postantibiotic era” of multidrug resistant “super-bugs.” Proposed politi-cal solutions, such as the Preservation of Antibiotics for Medical Treatment Act (PAMTA),would ban entire classes of subtherapeutic antibiotics (STAs) now used for disease preven-tion and growth promotion in food animals. The proposed bans are not driven by formalquantitative risk assessment (QRA), but by a perceived need for immediate action to preventpotential catastrophe. Similar fears led to STA phase-outs in Europe a decade ago. However,QRA and empirical data indicate that continued use of STAs in the United States has notharmed human health, and bans in Europe have not helped human health. The fears moti-vating PAMTA contrast with QRA estimates of vanishingly small risks. As a case study, ex-amining specific tetracycline uses and resistance patterns suggests that there is no significanthuman health hazard from continued use of tetracycline in food animals. Simple hypotheti-cal calculations suggest an unobservably small risk (between 0 and 1.75E-11 excess lifetimerisk of a tetracycline-resistant infection), based on the long history of tetracycline use in theUnited States without resistance-related treatment failures. QRAs for other STA uses in foodanimals also find that human health risks are vanishingly small. Whether such QRA calcula-tions will guide risk management policy for animal antibiotics in the United States remains tobe seen.
KEY WORDS: Acne; animal antibiotics; antimicrobial risk assessment; MRSA; Preservation of Antibi-otics for Medical Treatment Act (PAMTA); resistance risks; tetracycline
1. INTRODUCTION: STA REGULATIONSCIENCE AND POLITICS
This article discusses potential human healthrisks from continued use of subtherapeutic amountsof antibiotics to promote health and growth andprevent bacterial infections in food animals in theUnited States. Throughout, we use “subtherapeutic”
503 Franklin Street, Denver, CO, USA.∗Address correspondence to Tony Cox, 503 Franklin Street,
Denver, CO, USA; tel: 303-388-1778; fax: 303-388-0609; [email protected].
to refer to nontreatment-related doses; these mayhave health benefits for animals, but are not pre-scribed as therapy for treating diseases. The first partof the article, comprising Sections 1–7, discusses riskmanagement policy making with and without formalrisk assessment. We contrast risk management basedon an intuitively plausible narrative, suggesting thatfailing to stop animal antibiotic use now may imperilhuman health by promoting the spread of antibiotic-resistant “super-bugs,” with risk management basedon a more formal risk assessment approach thatsuggests that human health risks from this easily
432 0272-4332/10/0100-0432$22.00/1 C© 2010 Society for Risk Analysis
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 433
imagined scenario are probably quantitatively negli-gible. These styles pit a powerful intuitive and vis-ceral emotional response to a terrifying scenarioagainst the more reassuring results of highly cogni-tive and analytic data-driven calculations. Which willend up shaping U.S. policy remains to be determined.Sections 8–10 consider a specific, important example:hazard analysis of the continued use of tetracyclineantibiotics. This is a very widely used class of antibi-otics, and the one with the highest levels of resistancein multiple species of food-borne bacteria.
The current debate over subtherapeutic antibi-otics (STAs) illustrates a crucial question for thefuture role of risk analysis in public policy: Doesformal health risk analysis provide a trustworthy ba-sis for making key risk management policy deci-sions? How well can risk analysis succeed in guid-ing risk management policy making, especially whenintuition suggests easily imagined, vivid catastrophicconsequences (e.g., accelerated deterioration of theefficacy of life-saving antibiotics in human medicine)from failure to act promptly (e.g., by banning con-tinued use of STAs), but risk analysis indicates thatthe feared risks are actually trivial, and that recom-mended interventions are likely to do much moreharm than good? We see management of humanhealth risks from STAs as a test of whether formalrisk analysis, or more intuitive political approaches,will guide public health decision making when thetwo seem (at first) to conflict.
The following Sections 2–7 frame the currentdebate over continued STA use in terms of alter-native approaches to risk management—formal andanalytic versus intuitive. They compare U.S. andEuropean experiences, as illustrating risk analysis-driven and more intuitive, precautionary risk man-agement policy making, respectively. Sections 8–10focus on tetracyclines as an important class of an-tibiotics, identified by STA opponents as a prior-ity for removal, and consider sources of data andempirical evidence for assessing potential threats tohuman health from their continued use. The pur-pose is to compare risks of continued tetracyclineuse, as viewed from the perspective of formal riskanalysis (specifically, hazard identification based onmethodical consideration of specific uses and treat-ment failure data) with the same risks as perceivedfrom the perspective of a more holistic, intuitive as-sessment, based on a mental model for how contin-ued use might harm human health. These views maydrive quite different policy recommendations andpriorities.
2. POTENTIAL HUMAN HEALTH RISKSAND BENEFITS FROM STA USE
Subtherapeutic concentrations of antibioticshave long been used in feed and water to preventdisease and to promote health and growth in foodanimals. Such uses have at least two possible effectson human health (e.g., Phillips, 2007). On the onehand, conventional wisdom holds that healthy ani-mals make for healthy people. STAs improve animalhealth and reduce microbial loads in meats, therebypresumably reducing the potential for food-borne ill-nesses (Hurd et al., 2008; Arsenault et al., 2007; Singeret al., 2007). STAs can promote growth, particularlyin poultry and hogs, by improving nutrient absorp-tion and by depressing the growth of organisms thatcompete for nutrients, thereby increasing feed ef-ficiency (MacDonald and McBride, 2009). On theother hand, routine use of antibiotics in food ani-mals selects for antibiotic-resistant strains of bacte-ria in the GI tracts of food animals (e.g., Inglis et al.,2005). If these resistant bacteria were to spread inthe environment, or survive through the food chain,in sufficient concentrations to cause food poisoningwhen people eat raw or undercooked meats, and ifthe resulting illnesses were then treated with antibi-otics to which the bacteria are resistant, then treat-ment might be less effective than if the infectingstrains of bacteria were more susceptible to the pre-scribed antibiotics (Slaughter, 2008). This frighten-ing scenario has great narrative plausibility; it stirsthe imagination. Whether it could occur in realitydepends on numerous prosaic details of food pro-cessing, food preparation, microbiology, and clini-cal practice, such as whether bacteria survive thetrip from farm to fork; whether physicians screen forresistance before prescribing an antibiotic; whetherprescribed doses are sufficient to kill even resistantbacteria; and whether hospitals serve raw or under-cooked meat to severely immunocompromised pa-tients in intensive care units.
Whether the human health benefits are greateror less than the human health harms from STA usehas seldom been quantified. Rather, the possibilityof harm is often pointed to as a sufficient reason toban STAs, without further analysis or comparisons(ibid). Europe has accepted this view, and accord-ingly has phased out STAs; the United States has notdone so in the decade since the EU bans, however.The rationales for these different risk managementresponses, and subsequent public health experiencesin the United States and Europe, are discussed next.
434 Cox and Popken
3. EUROPEAN AND U.S. REACTIONS TOPRESSURES TO BAN STA USE
The usual medical consequence of reducedeffectiveness of antibiotic therapy for food-borne ill-nesses is extra hours or days of diarrhea (e.g., FDA-CVM, 2001). In severely ill patients with compro-mised immune systems (such as AIDS, leukemia,or organ transplant patients), however, antibiotic-resistant zoonotic infections can be life-threatening.For decades, envisioning hypothetical scenarios inwhich animal antibiotic use increases the rate of an-tibiotic treatment failures in desperately ill humanpatients has spurred many activists, scientists, physi-cians, and some politicians to repeatedly call for banson the use of STAs in food animals.
In response, starting in the late 1990s, the Euro-pean Union phased out growth promoting STA use,invoking the precautionary principle as justification,despite the recommendations of scientific reviewerswho pointed out that STA use has no apparent causallink to resistance rates in human infections in thereal world (Pugh, 2002). The United States has notyet done so, largely because its regulators have re-fused to withdraw products that benefit animals (and,thereby, perhaps human health) when empirical datado not show that they harm human health.
For many regulators in the United States, ini-tial narrative plausibility and activist outrage do not(as of this writing) provide adequate bases for regu-lation. They insist that regulatory action should re-flect verifiable empirical data and risk assessment,showing that advocated interventions address realproblems (supported by empirically valid cause-and-effect relations, rather than only by envisionedscenarios), and that they are likely to create real pub-lic health benefits (rather than no change, or unin-tended public health harm). But political pressure onCongress remains strong to eliminate STAs, withoutfurther empirical research or quantitative risk assess-ment. This pressure comes in part from concernedactivist groups, such as Keep Antibiotics Working(KAW, 2009); in part from concerned scientists andphysicians (UCS, 2009; Simmons, 2007); and in partfrom powerful politicians in Congress (Pew, 2009;Simmons, 2007).
The possibility of using political concern, ratherthan probable health consequences, as a basis forSTA regulation has been strengthened by the 2003release of FDA’s Guidance Document for Industry#152 (FDA-CVM, 2003). This document allows qual-itative expressions of concern to be mapped to rec-
ommended levels of risk management priority, withno need to consider how or whether recommendedinterventions would change human or animal healthrisks. Mathematical analysis shows that this quali-tative approach can ignore information that is logi-cally necessary to correctly compare risks, leading toworse-than-random risk management decisions andresource allocations (Cox et al., 2005; Cox, 2008).Guidance Document #152 does not seek to eval-uate the changes in human health that would becaused by withdrawing STAs, but instead provides aframework for documenting levels of concern (e.g.,as “High,” “Medium,” or “Low,” based on factorssuch as the judged “importance” level (e.g., “VeryImportant” or “Critically Important”) of classes ofantbiotics for human medicine. Such concern-basedrisk management frameworks can support actionsthat exacerbate the problems they mean to rem-edy (Cox, 2007). In response to Guidance Docu-ment #152, several animal pharmaceutical companieshave taken the initiative to supplement their reviewsubmissions with additional quantitative data or riskanalyses.
4. EVIDENCE OF CAUSALITY:ATTRIBUTION ASSUMPTIONSVERSUS EMPIRICAL DATA
Pressure to phase out STAs in the United Statesis buttressed by a widespread and much-repeatedbelief—presented as fact in many peer-reviewed sci-entific journal articles, as well as in less formalsources—that STA use in food animals has beencausally linked, via consumption of resistant bacte-ria transmitted via the food chain, to increased ratesof real-world treatment failures and harm in humans(e.g., Mølbak, 2004; Slaughter, 2008). However, webelieve that this ubiquitous impression is mistaken—that it is only a scientific urban legend. It is basedlargely on epidemiological studies that misinterpretregression coefficients or odds ratios in multivari-ate statistical risk models, between food consump-tion patterns and resistant illnesses, as evidence ofcausal relations (e.g., Angulo et al., 2004; Varmaet al., 2006); such causal interpretations are in generalnot justified (Cox, 2005). To our knowledge, no caseof a treatment failure in a human patient, caused bytransmission of antibiotic-resistant bacteria throughthe food chain (“from farm to fork”), has ever beendocumented in the United States. Nor have decadesof efforts by groups such as the Alliance for Pru-dent Use of Antibiotics, as well as many regulatory
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 435
scientists, microbiologists, and physicians, succeededin showing that STA use has had any adverse im-pact on human health via food-borne transmissionof resistant bacteria. Rather, it has become commonto simply assume, and then to assert, that STA useharms human health. This is sometimes done by firstestimating “attributable fractions” that assign some(perhaps subjectively estimated or hypothetical) pro-portion of human treatment failures to STAs (e.g.,Barza and Travers, 2002) and then treating this as-sumption as if it were valid empirical data or evi-dence (e.g., Angulo et al., 2004). Regulators, activists,and politicians routinely refer to such assumptions,estimates, and attributions as “evidence” of a causalrelationship between STA use and increased humanhealth harm, but they are not. For the same methodscan be used equally well (or badly) to create a posi-tive statistical relation between any two positive ran-dom variables, even if they are completely unrelated(statistically independent), or are negatively related(Cox, 2005). Such force-fit “evidence” is not evidenceat all.
5. A POLITICAL APPROACH TO RISKMANAGEMENT WITHOUTRISK ANALYSIS
The reasoning and rhetoric now being used tourge Congress to halt STAs use in the United Statesare well illustrated in the following congressional tes-timony (Slaughter, 2008):
We cannot in good conscience stand by while our life-saving antibiotics become obsolete. While overuse ofantibiotics among humans is certainly a major causefor increasing resistance, there is evidence that thewidespread nontherapeutic use of antibiotics in animalfeed is another cause of heightened resistance. A Na-tional Academy of Sciences report states that, “a de-crease in antimicrobial use in human medicine alonewill have little effect on the current situation. Sub-stantial efforts must be made to decrease inappropri-ate overuse in animals and agriculture as well.” . . . Thenontherapeutic use of antibiotics in poultry skyrocketedfrom 2 million pounds in 1985 to 10.5 million pounds inthe late 1990s. This kind of habitual, nontherapuetic useof antibiotics has been conclusively linked to a grow-ing number of incidents of antimicrobial-resistant in-fections in humans, and may be contaminating groundwater with resistant bacteria in rural areas. . . . Duringdiscussions involving the now-enacted Farm Bill, I sup-ported language which would have provided the farmindustry with sound, scientific information on produc-tion practices that could have helped them reduce theirdependence on antibiotics and meet the growing con-sumer demand for meat produced without these drugs.
The ability to grow food animals with fewer antibi-otics would have also given US exporters an advantagein the international marketplace. . . . Disappointingly,however, industry successfully lobbied to strip this lan-guage out of the Farm Bill. . . . I am also the sponsor ofH.R. 962, the Preservation of Antibiotics for MedicalTreatment Act (PAMTA). This bill requires three ac-tions to accomplish the goal of reducing antibiotic resis-tance in humans. PAMTA would phase out the use ofthe seven classes of medically significant antibiotics thatare currently approved for nontherapeutic use in animalagriculture. . . . [I] cannot stress the urgency of this prob-lem enough. When we go to the grocery store to pick updinner, we should be able to buy our food without wor-rying that eating it will expose our family to potentiallydeadly bacteria that will no longer respond to our med-ical treatments.
From the standpoint of sound risk analysis and riskmanagement policy making, such impassioned state-ments are striking for several reasons. First, there isno apparent examination of whether discontinuingSTA use will actually help human health, or harmit. There is no discussion of whether improved foodsafety from reduced bacterial loads in meats, or in-creased risks of antibiotic resistance among bacteriathat survive, dominates the public health impacts ofSTA use. Instead, the statement moves directly froma problem statement (antibiotic resistance in humanbacterial infections threatens human health) to a rec-ommended solution (discontinue STA use in food an-imals), by way of an assumption (that eating meat willexpose families to “potentially deadly bacteria thatwill no longer respond to our medical treatments”).There is no attempt to apply the usual intermedi-ate risk analysis steps of quantifying the claimed riskand explicitly assessing, documenting, and compar-ing the probable consequences of risk managementalternatives, such as continuing versus discontinuingprudent use of STAs to promote health. Yet, theseomitted steps are the heart of rational risk analysis.
5.1. Quantitative Versus Qualitative Descriptionsof Risks
Second, the above calls for urgent action do notquantify the hoped-for reduction in risk that theiradvocates presumably anticipate will result from theproposed inteventions. The dreaded nature of the en-visioned scenario (families around their dinner ta-bles being exposed to untreatable, deadly bacteria) isvividly expressed, but the frequency and severity ofresulting preventable adverse health consequences,if any, are not described. When this is done, theresulting numbers turn out to be vanishingly small,
436 Cox and Popken
robbing the dramatic narrative and urgent calls forprompt action of much of their visceral appeal. Forexample:
• For macrolides, Hurd and Malladi (2008) con-cluded that “the predicted risk of suboptimalhuman treatment of infection with C. coli fromswine is only 1 in 82 million; with a 95% chanceit could be as high as 1 in 49 million. Risks fromC. jejuni in poultry or beef are even less.”
• For penicillin, Cox et al. (2009) calculated that“not more than 0.037 excess mortalities peryear (under conservative assumptions) to 0.18excess mortalities per year (under very con-servative assumptions) might be prevented inthe whole U.S. population if current use ofpenicillin drugs in food animals were discon-tinued, and if this successfully reduced theprevalence of antibiotic-resistant E. faeciuminfections among intensive care unit (ICU) pa-tients.” They note that the true risk could wellbe zero.
• For streptogramins, banning virginiamycin hasbeen estimated to prevent from 0 to less than0.06 statistical mortalities per year in the en-tire U.S. population (Cox and Popken, 2004;see also FDA-CVM, 2004).
Such specific, small risk numbers may be less likelyto incite congressional intervention than the broadqualitative assertion that “[w]e should be able tobuy our food without worrying that eating it willexpose our family to potentially deadly bacteriathat will no longer respond to our medical treat-ments.” In fact, the “deadly” bacteria referred to,including commensals such as MRSA, Campylobac-ter, and E. faecium, are routinely found—and areharmless—in healthy people (and other birds andmammals) with normally functioning immune sys-tems. They become life-threatening only under veryunusual and rare circumstances, especially for pa-tients with severely compromised immune systems,such as AIDS, leukemia, and transplant patients, typ-ically in ICUs. Attributing the risks from such rare,dire, medical circumstances to consumption of meatat family meals is perhaps effective rhetoric, but it ispoor microbiology. Such qualitative, concern-drivenrisk management, stripped of relevant risk numbers,data, alternatives, causation, and consequence esti-mates, may motivate political support for interven-tions, but it can also encourage different, less effec-tive, risk management recommendations than wouldresult from sound QRA (Cox, 2007).
5.2. Data-Driven Hazard Identification VersusUnverified Causal Assertions
A third aspect of the above statements on therationale for phasing out STAs in the United Statesalso differs sharply from the typical, more cautious,risk analysis presentation of verifiable facts, data,and models to support specific quantitative conclu-sions. This is their reliance on sweeping qualitativecausal claims: specifically, that “[a] decrease in an-timicrobial use in human medicine alone will havelittle effect on the current situation,” (which ap-pears to be a misquote of the much weaker state-ment that “[c]learly, a decrease in the inappropriateuse of antimicrobials in human medicine alone is notenough,” http://books.nap.edu/openbook.php?isbn=030908864X&page=207) and that “[t]he ability togrow food animals with fewer antibiotics would havealso given US exporters an advantage in the inter-national marketplace.” Hazard identification—thepart of risk assessment that traditionally considersevidence for (and against) hypothesized risks andexposure-response relations, and the empirical basesfor claimed adverse effects—shows that these claimsare not well supported by data.
The assertion that banning STAs would cre-ate “an advantage in the international marketplace,”made without reference to any specific supportingdata, contrasts strongly with the historical experi-ences of countries that have tried it. For example,Sweden banned STAs in 1986. “The Swedish expe-rience shows that antibiotics are not necessary toproduce healthy animals, provided their living con-ditions, rearing and foods are improved. This didcome at a cost: thousands of pigs and chickens prob-ably died as a direct result of the ban, despite theoverall improvement in animal welfare. Swedish pro-duce is more expensive, and so less competitive onthe market, and the costs of the venture are expen-sive” (Hughes and Heritage, 2004, emphasis added).Norway and the United Kingdom had similar experi-ences. The problem was great enough in those coun-tries that alternative, prescription antibiotics were in-troduced in response to significant enteric diseases inpoultry (Van Immerseel et al., 2009).
The crucial, policy-relevant generalization that“a decrease in antimicrobial use in human medicinealone will have little effect on the current sit-uation” is flatly contradicted by many quantita-tive studies over many years. For example, morethan a decade ago, there were already “more than20 studies on consistent associations, dose-effect
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 437
Fig. 1. MRSA incidence in human hospital patients reflects use ofhuman antibiotics.Source: Monnet et al., 2004.
relationships, and concomitant variations, all sup-porting a causal relationship between [human pa-tient] antimicrobial-drug use and MRSA” (Monnetand Frimodt-Moller, 2001). Gould (1999) noted that:“Early studies in various hospitals showed rapid re-versal of major clinical problems of resistance tochloramphenicol, erythromycin and tetracycline inStaphylococcus aureus on withdrawal of these antibi-otics from clinical use.” Moreover, “[i]n the com-munity, outbreaks of erythromycin-resistant GroupA streptococci and penicillin-resistant pneumococcihave been controlled by major reductions in prescrib-ing of erythromycin and penicillin.” They concludethat “there is little doubt that careful antibiotic pre-scribing can curtail the emergence and reduce theprevalence of resistance” in these and other bacteria.
Similarly, Aldeyab et al. (2008) studied in de-tail the variance in monthly incidence of Methicillin-resistant Staphylococcus aureus (MRSA) over a5-year period in one hospital. They found that “tem-poral variations in MRSA incidence followed tempo-ral variations in the [human] use of fluoroquinolones,third-generation cephalosporins, macrolides andamoxicillin/clavulanic acid (coefficients = 0.005, 0.03,0.002 and 0.003, respectively, with various timelags).” Over three-quarters of the monthly variancein MRSA was explained by a model that includedonly these human antibiotic usage variables (whichincrease MRSA prevalence) and infection controls(which reduce it). Fig. 1 shows results from a simi-lar 2004 study. Again, decreasing antimicrobial usein human medicine alone dramatically reduced sub-sequent MRSA rates. A time series model of MRSAchanges over time, with only human antibiotic use asexogenous explanatory variables, accounts for over90% of the observed variance in MRSA rates. In
short, not only is it untrue that “a decrease in antimi-crobial use in human medicine alone will have littleeffect on the current situation” (Slaughter, 2008), butin many studies, over many years, it is almost the onlything that does have an effect.
Other time series analyses have revealed that,instead of MRSA first increasing in the commu-nity (e.g., from families eating contaminated meat)and then infiltrating hospitals as sick patients enter,the flow is in the opposite direction: “[W]e demon-strated that variations in MRSA prevalence in thehospital are quickly followed by similar variationsin MRSA prevalence in the surrounding community.These results suggest that the reason for the increasein MRSA prevalence in the community was a hospi-tal MRSA outbreak” (MacKenzie et al., 2007).
Traditional risk analysis typically avoids makingsweeping generalizations about causal relations (suchas that “a decrease in antimicrobial use in humanmedicine alone will have little effect”), in favor of dis-playing relevant data supporting more specific, quan-titative, causal claims. This more cautious, empiri-cally driven, approach tries to reduce the likelihoodthat decision makers will adopt wholly false beliefsabout what works, and hence will urge ineffectiveor harmful policies based on such mistaken beliefs.Hence, risk analysts typically put hazard identifica-tion early in a risk assessment, insisting that empiricalevidence of causal relations be documented and scru-tinized before undertaking further quantification ofexposure-risk relations. In the present example of re-sistance risks, a well-conducted hazard identificationwould consider time series studies and data showingthat changes in human antibiotic use precede, and ex-plain, almost all of the subsequent changes in antibi-otic resistance in isolates from humans (MacKenzieet al., 2007). It would examine time series data show-ing that changes in infections in humans tend to pre-cede, rather than to follow, corresponding changes inCampylobacter prevalence in chicken flocks (Chris-tensen et al., 2001). It would consider evidence thatresistant bacteria are transmitted from humans tofood animals, and from environmental sources (e.g.,flies, soil, water) to both, and would consider theextent to which these flows might explain instancesof similar or identical bacteria in different species(e.g., Herron-Olson et al., 2007). Such relevant em-pirical data would be evaluated, critically discussed,and synthesized in the hazard identification sectionof a typical risk assessment before assuming (or as-serting) that changes in animal consumption of STAsdrive changes in resistance to infections in hospital-ized patients.
438 Cox and Popken
Policies based on inadequate understanding ofcausal relations can produce the opposite of their in-tended effects, e.g., by inducing shifts in antibioticuse that unintentionally increase costs and harm tohuman health (Beilby et al., 2002), or that increaseanimal use of therapeutic antibiotics that threatenhuman health more than the STAs that they re-place (Casewell et al., 2003). If decisionmakers aremistakenly told that reducing STA use will sig-nificantly reduce resistant infections in human pa-tients, but that reducing human medical antibioticuse alone would make little difference, then inter-ventions based on such assumptions may not achievetheir intended public health goals. Risk manage-ment policies based on false causal premises—nomatter how well intended, or how sincerely andpassionately urged—cannot be expected to causedesired changes. Risk analysis, unlike many other po-litical and precautionary approaches, therefore fo-cuses on understanding—and, when possible, evenquantifying—the probable causal effects of differentinterventions on outcomes of concern, prior to rec-ommending risk management interventions. Correct,usefully detailed, causal understanding and quantita-tive modeling of effects and tradeoffs may be essen-tial for effective intervention in systems as complexand difficult to control as antibiotic resistance in hu-man infections. Otherwise, well-intended, common-sense interventions may backfire, producing the op-posite of their intended effects (e.g., Beilby et al.,2002).
6. EMPIRICAL EVIDENCE FROM THEUNITED STATES AND EUROPE: DO STAsHARM OR HELP HUMAN HEALTH?
Empirically, how has STA use affected humanhealth in the United States? Conversely, how haswithdrawal of STAs affected human health in Eu-rope? Although unambiguous causal interpretationand explanation of historical trends is admittedlychallenging, and any such interpretation of resistancetrends is at present only hypothetical, the above tes-timony (Slaughter, 2008) emphasizes that nonther-apeutic use of antibiotics in poultry had “skyrock-eted” in the United States by the late 1990s—ahistory that should help to reveal the human healthconsequences (if any) of rapidly increasing STA use.Continued use of STAs in the United States wasfollowed by improvements in microbial safety andzoonosis-related human health. For example, micro-bial loads of campylobacter in poultry dropped to
perhaps 10% of their previous levels by 1995, whilehuman campylobacteriosis health risks decreased sig-nificantly (Stern and Robach, 2003). Use of animalantibiotics in the United States coincided with re-duced risks of many other food-borne illnesses, too:“In comparison with 1996–1998, relative rates ofYersinia decreased 49% (CI = 36%–59%), Listeriadecreased 42% (CI = 28%–54%), Shigella decreased36% (CI = 9%–55%), Campylobacter decreased31% (CI = 25%–36%), STEC O157 decreased 25%(CI = 9%–38%), and Salmonella decreased 8%(CI = 1%–14%) in 2007” (CDC, 2008). These de-clines in food-borne illness rates were largely accom-plished by 2004, and have generally persisted since,while most antibiotic resistance rates in zoonotic bac-teria have remained stable or declined (Cox andRicci, 2008; NARMS, 2009).
Even as the United States was enjoying fallinglevels of campylobacteriosis and other food-borne ill-nesses, multiple European countries that had ceasedusing STAs experienced significant increases inhuman illnesses due to food poisoning, includingcampylobacteriosis (Vierikko et al., 2004) and otherzoonotic bacteria. Increased human illnesses wereaccompanied by greater human use of antibioticsand, ironically, by jumps of up to several hundredpercent in some antibiotic resistance rates in hu-man patients (Hayes and Jensen, 2003). Most ofthese changes were complete by about 2004, and ex-tensive recent interventions to control Salmonellaand Campylobacter have achieved some successes(Cox, 2007). Nonetheless, a decade after the bans,Denmark, a long-time leader in implementing andadvocating bans on STAs, and in collecting data tomonitor the results, reported that mean hospital con-sumption of antibiotics had increased by 63% be-tween 1997 and 2007, leading some reviewers to notethat: “Overall, Denmark’s position as a country char-acterised by rational use of antibiotics and a lowoccurrence of resistance is under pressure, and ini-tiatives to counter such tendency are needed” (EPI-NEWS, 2008).
It is notoriously difficult to draw valid causal in-ferences from such aggregate (“ecological”) longitu-dinal observational data alone, and the logical fal-lacy of inferring causation from temporal sequence(post hoc, ergo propter hoc) must be avoided. Validcausal inference requires more detailed causal analy-ses and models, such as the time series analyses dis-cussed above (see Fig. 1). However, it is clear thatany hopes that withdrawing STAs in Europe wouldcause substantial reductions in serious antibiotic
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 439
resistance problems in human patients have not beenwell supported by historical experience.
7. PREVIOUS HUMAN HEALTHRISK-BENEFIT COMPARISONS FOR STAs
Europe’s acceptance of the precautionary prin-ciple encourages policymakers to implement well-intended, popular policy interventions, even if theirtrue human health (and other) consequences cannotyet be predicted. In the United States, by contrast,regulators have been less willing to experiment withdiscontinuing STAs than in Europe. Predictions ofhow withdrawing STAs would probably affect hu-man health in the United States must therefore bebased on risk assessment models, rather than on his-torical experience.
Relatively few studies have explicitly comparedthe human health benefits to the potential humanhealth risks of continued STA use in the UnitedStates. Studies that have made this comparison havenot generally favored discontinuing use. For exam-ple, for virginiamycin, potential human health ben-efits from illnesses prevented have been estimatedto be more than 10,000 times greater than poten-tial human health risks from increased resistance(which are estimated to be less than 1 excess treat-ment failure-related mortality per decade in the U.S.population) (Cox and Popken, 2004). For macrolides,human health benefits from continued use (due tohuman illnesses prevented by safer food) have beenestimated to exceed the human health costs, due tohuman illnesses prolonged by resistance, by a ratio ofmore than 1,000:1 (Cox and Popken, 2006). For otherSTAs, direct comparisons of human health benefitsand costs are not yet available.
8. ASSESSMENT OF HUMAN HEALTHHAZARDS FROM TETRACYCLINERESISTANCE
It is easy to extol the virtues of proper hazardidentification as a prelude to risk assessment andrisk management policy making, but useful hazardidentification can be challenging when the specificexposures, harms, and pathways of interest are un-known or are very uncertain. For example, it hasbeen hypothesized that using tetracyclines in foodanimals might help to select for bacteria, such asMRSA, E. coli, or multidrug resistant (MDR) strainsof Salmonella that also resist other antibiotics (e.g.,Akwar et al., 2008) and this co-selection might cre-
ate risks larger than those from resistance to tetra-cyclines per se. Past risk analyses of animal-useantibiotics have generally focused on ones withhuman-use counterparts used to treat illness causedby food-borne pathogens such as E. coli, Campy-lobacter, Salmonella, and Enterococci (FDA-CVM,2001; Cox and Popken, 2002, 2004; FDA-CVM, 2004;Alban et al., 2008; Hurd et al., 2008; Cox et al., 2009).By contrast, tetracyclines, while heavily used in bothanimals and humans, are not typically used to treatfood-borne diseases. Therefore, hazard identificationmust consider how they are used, as well as evidenceabout co-selection hazards (i.e., the potential for bac-teria that are resistant to one drug to be selected byexposure to another). A literature search identifiedMRSA and multidrug resistant (MDR) Salmonellaas possible co-selection concerns; they are examinednext.
8.1. Assessment of Potential Hazard fromCo-Selection of MRSA by Tetracycline STAs
Recent research in Denmark (Lewis et al., 2008)has identified a specific strain of MRSA, CC 398 (alsocalled ST 398), that is associated with pigs, but that isincreasingly being found in humans living or work-ing on a farm with animals. MRSA is usually a harm-less commensal; thus, the finding of the same bacte-ria in humans and pigs does not by itself indicate ahealth hazard. (Indeed, as noted by Morgan (2008),one might equally well hypothesize that humans arecausing the spread of MRSA in pigs as the reverse;in neither case does carriage per se imply infection orhealth harm.) However, the emergence and spreadof community-acquired MRSA infections is a grow-ing concern in several countries, including the UnitedStates.
In the United States, Smith et al. (2008a) re-ported that 45% of farm workers and 70% of swinesampled on seven farms that were part of one closedsystem in east Iowa and west Illinois tested positivefor MRSA. (Subsequent sampling (Smith, 2008b)brought the overall average down to 49% of swine,and confirmed the strain as ST398.) Another sam-pled establishment, in contrast, had no MRSA inpigs or workers. All MRSA samples were resistant topenicillin, oxacillin, and tetracycline. In a recent in-terview (Schneider, 2008), Dr. Smith suggested thatanimal use of tetracyclines was the “cause of thespread of [human] MRSA.” However, in a subse-quent peer-reviewed publication (Smith et al., 2009),the authors carefully noted that “both production
440 Cox and Popken
systems that we sampled employ similar protocolsfor prophylactic and therapeutic use of antimicrobialagents, including tetracycline. Therefore, our datado not allow us to speculate on the relationship be-tween antimicrobial use and MRSA carriage.” NoMRSA types commonly found in human infections inNorth America (USA100, USA300, USA400) werereported. Politicians (Slaughter, 2008) and activists(KAW, 2008) have implicated animal antibiotic usein general as a cause of spreading MRSA, and haveadvocated curtailing all “inappropriate” (nonthera-peutic) use of antibiotics in agriculture.
However, to go from the unsurprising observa-tion that some humans that work around pigs carrythe same strain of MRSA as pigs to the inference thatanimal uses of antibiotics [including tetracyclines]cause the spread of MRSA in human patients requiresa leap across many unestablished causal links. For ex-ample, there is no evidence that tetracycline use inanimals increases the spread or prevalence of animalMRSA, nor that MRSA in farm animals is conveyedto consumers on retail meats, nor that MRSA con-veyed via meats could cause infections in consumers.Recent studies in the United States showed MRSArates of 0% (Chan et al., 2008), 3% (Vedder, 2008),and 5% (Pu et al., 2009) in retail meat. DNA straintesting was only performed in the latter study, whichfound that none of the isolates was of type ST398.Therefore, no ST 398 strains have yet been found onretail meats in the United States.
A Dutch study of retail meats (van Loo et al.,2007) found 2 MRSA strains in 79 samples (2.4%),one of which was ST398, but that was present in “verylow amounts” “not likely to cause disease.” Further,there have been no reports of MRSA ST398 amonghumans, other than farm workers, in the UnitedStates. Most strains are USA100 (health-acquired),or USA300 (community-acquired) (Klevens, 2007).The scenario in which MRSA ST 398 infectsconsumers via the food chain lacks empiricalsupport.
Tetracycline resistance observed in MRSA doesnot, by itself, pose a health risk. Doxycycline isFDA-approved for the treatment of S. aureus skininfections, but not specifically for those caused byMRSA (CDC, 2006). Although international bod-ies such as WHO have sometimes classified tetracy-clines as “Critically Important Antimicrobials,” onthe grounds that they are “a limited therapy for in-fections due to MDR [multidrug resistant] S. au-reus,” this appears to reflect a failure to distinguishappropriately between a relatively new drug, tigecy-
cline, and older tetracycline drugs. In reality: “Theonly tetracycline in the WHO critically importantlist is a glycylcycline (tigecycline), the other classmembers are categorized as highly important. Dueits resistance mechanisms, tigecycline is regarded asrepresenting a different generation to other tetracy-clines” (FAO/WHO/OIE, 2008, p. 31). Tigecycline,which was approved by the FDA in 2005 specificallyto treat MRSA, is not used in food animals and itdoes not have the same resistance mechanisms orprofile as tetracycline. Thus, it would be a mistake(a logical fallacy of composition) to classify tetracy-cline per se, or tetracycline drugs used as STAs, as“critically important” for human medicine, based ontigecycline.
Would banning STAs reduce the risk or speedof MRSA emergence? European experience suggestsnot. Concerns about the spread of MRSA from foodanimals to humans began in Europe, with the identi-fication of pigs as a reservoir for human clonal com-plex (CC) 398 MRSA, in the Netherlands, France,Denmark, and later, Canada (Lewis et al., 2008).MRSA has developed and spread in these coun-tries, undeterred by the fact that most growth pro-motion STAs (including tetracyclines and penicillins)have been banned there for many years. The ob-served easy spread of CC 398 MRSA among pigs andveterinarians (and other humans) in Europe (Wulfet al., 2008) suggests that the growth promotionSTA bans do not prevent or noticeably inhibit thissource of community-acquired MRSA. Tetracyclineresistance (along with ciprofloxacin, erythromycin,and mupirocin resistance) is also commonly presentin human strain USA300, which originated in ur-ban settings in the early 2000s (Russell, 2008). Thisstrain did not spread because of tetracycline expo-sures in agricultural settings, but because of humanuse. Moreover, as already noted in Fig. 1 and manyother studies, MRSA in human patients overwhelm-ingly follows human use of antibiotics, especiallyquinolones (e.g., Aldeyab et al., 2008; Bosso andMauldin, 2006; Tacconelli et al., 2008; Weber et al.,2003). Thus, we conclude that there is no evidencethat STA use, including tetracycline use, contributesto the emergence and spread of either community-acquired or hospital-acquired MRSA infections in hu-man patients. Although the hypothesis that STA usecontributes to MRSA risk in human patients is per-haps intuitively plausible a priori, data do not sup-port an inference that banning STAs would produce(or, in Europe, has produced) any detectable reduc-tion in MRSA risks. But data do clearly indicate
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 441
that reducing use of quinolones and third-generationcephalosporins in human medicine reduces MRSArisks.
8.2. Assessment of Potential Hazard fromCo-Selection of MDR Salmonellaby Tetracycline
The Salmonella species of most concern in thetransmission of disease from animals to humans areSalmonella enteriditis and Samonella typhimurium(WHO, 2005). These species cause gastroenteritisin humans, a condition that is often uncomplicatedand does not require treatment, but may cause se-vere illness or even death in the young, elderly, orimmunocompromised. For those needing treatment,fluoroquinolones are the drug of choice for adults,while third-generation cephalosporins are mostoften used for children or those who cannot toler-ate fluoroquinolones. To a lesser extent, chloram-phenicol, ampicillin, amoxicillin, and trimethoprim-sulfamethoxazole are sometimes used (WHO, 2005);therefore, tetracyclines are not used in treatingSalmonella.
Recent years have seen the development of vari-ous strains of multidrug resistant (MDR) Salmonellathat are resistant to a variety of antimicrobials,including fluoroquinolones and cephalosporins. Inthis section, we examine the extent to which animaluse of tetracyline causes or encourages the devel-opment of MDR Salmonella, particularly in swine,where tetracycline use is thought to be the mostintense. For example, “the increasing significanceof swine as reservoirs of emerging MDR serovars”has been identified as a recent public health concern(e.g., Patchanee et al., 2008). Common multidrugresistance patterns seen in swine include amoxi-cillin/clavulanic acid-ampicillin-chloramphenicol-piperacillin-tetracycline (Gebreyes et al., 2004);streptomycin, sulfamethoxazole, and tetracycline;and ampicillin, chloramphenicol, streptomycin,sulfamethoxazole, and tetracycline [ACSSuT](Gebreyes et al., 2006). The latter combination istypical of Salmonella serotype Typhimurium DT104,a species of worldwide concern. Of particular con-cern is the recent development of a penta-resistantform of Salmonella, serotype Newport MDR-AmpC,that is resistant to all of the ACSSuT antimicrobialsas well as amoxicillin-clavulanic acid, cephalothin,cefoxitin, and ceftiofur, and exhibits decreased sus-ceptibility to ceftriaxon. Note this includes second-
and third-generation cephalosporins, the latter aSalmonella first-line treatment.
There is evidence that animal use of tetracyclinesdoes not drive the development of MDR Salmonella,specifically including cephalosporin-resistant vari-eties. First, cephalosporin-resistant varieties of MDRSalmonella did not appear in the United States un-til the late 1990s, despite extensive use of tetra-cyclines in animal production in the United Statessince the 1950s, suggesting that this extensive use didnot by itself select for cephalosporin cross-resistance.For example, Berge et al. (2004) examined samplesof Salmonella enterica subspecies enterica serovarNewport from humans and animals taken in twotime periods: 1988–1995 and 1999–2001. PFGE anal-ysis showed that, while the genetic makeups ofthese two groups were similar, only the more re-cent samples were resistant to cephalosporins. A na-tionwide surveillance study of Ceftriaxone-resistantSalmonella infections in humans in the United Stateswas undertaken by Dunne et al. (2000). The preva-lence of ceftriaxone-resistant Salmonella was 0.1% (1of 1,326) in 1996, 0.4% (5 of 1,301) in 1997, and 0.5%(7 of 1,466) in 1998. A study of multidrug-resistant(MDR) Salmonella Typhimurium in humans, retailmeat, and food animals from Yucatan, Mexico wasperformed by Zaidi et al. (2007). MDR Salmonellatyphimurium containing the blaCMY-2 gene (confer-ring cephalosporin resistance) rose from 0% (0/27)during 2000 and 2001 to 75% (63/84) in 2004 and2005. It is apparent that widespread dissemination ofcephalosporin-resistant Salmonella in North Amer-ica is relatively recent. While this does not necessarilymean that tetracyclines play no role in the develop-ment of cephalosporin-resistant Salmonella, clearlyother factors must be at work.
Secondly, Cephalosporin-resistant Salmonellaactually appear to be relatively rare among swine. Astudy by Weigel and Isaacson (2004) analyzed 207Salmonella isolates taken from swine and environ-mental sources on 11 swine farms in Illinois. Tenof the 11 farms used tetracycline as a feed addi-tive, but at varying rates. Resistance was tested tocephalosporin drugs, including cephalothin (1st gen-eration), cefoxitin (2nd generation), ceftiofur (3rdgeneration), and ceftriaxone (3rd generation). Re-sistance was detected only for the first-generationcephalosporin, cephalothin, which had a total of6 positive samples from two of the 11 farms witha total resistance prevalence rate of 6.8%. Thisstudy also showed only a weak association be-tween antibiotic usage levels and prevalence of
442 Cox and Popken
Salmonella resistant to the same antibiotic. Gray et al.(2004) performed a large study of 5,709 Salmonellaenterica isolates from a variety of animal species inthe United States. A total of 112 isolates were resis-tant to ceftiofur and ceftriaxone and also possessedthe blaCMY gene. Ten of these were from the 1,580swine isolates (0.6% resistance rate). Much higher re-sistance rates were found in isolates from other ani-mals, as follows: horses (9.5%), cats (8.3%), turkeys(4.7%), dogs (4%), cattle (1.9%), and chickens (1%).Similarly, a large nationwide study by Varma et al.(2006) of Salmonella serotype Newport-MDRAmpC(which is cephalosporin-resistant) in humans foundstrong associations with previous use of antibiotics,and consumption of raw ground beef or runny eggs.They concluded that the infections are likely ac-quired through food with bovine or perhaps poultrysources.
We also examined evidence regarding whethertetracycline use may be responsible for the earlierseen forms of MDR Salmonella, dangerous in them-selves, but that could also theoretically provide astepping stone to the cephalosporin-resistant vari-eties discussed above. To examine this potential haz-ard empirically, Gebreyes et al. (2006, 2008) and col-leagues (Thakur et al., 2007) compared antimicrobialresistance rates in 60 conventional and antimicrobial-free (ABF) swine production systems in threestates. They found that Salmonella prevalence wassignificantly higher (by a factor of almost four-fold, 15.2% vs. 4.2%) among the antimicrobial-freesystems than in the conventional systems. A signif-icant proportion of the isolates from antimicrobial-free herds were resistant to a variety of antimi-crobial agents. A multidrug resistance pattern withresistance to streptomycin, sulfamethoxazole, andtetracycline was commonly observed, but there wasno significant difference in the proportion of isolateswith this pattern between the conventional (19.5%)and the antimicrobial-free (18%) systems. TheACSSuT multiresistance pattern was also commonin the antimicrobial-free herds. The authors interpretthese data as “suggesting selective pressure otherthan antimicrobial use, could be important risk fac-tors for the persistence of MDR Salmonella strainsin the swine production environment” (Gebreyes,2008).
Other researchers have examined genetic evi-dence and arrived at similar conclusions, that “an-timicrobial selection pressure does not consistentlyexplain the increased prevalence of epidemic MDRstains of S. enterica, and restricting antimicrobial use
often fails to control the dissemination of epidemicMDR strains, suggesting that there may be other bio-logical traits or genetic factors that increase bacterialvirulence or fitness or at least compensate for a fit-ness cost mostly accompanied by antimicrobial resis-tance” (Kang et al., 2006). While these selective pres-sures are not fully understood, possible explanationsare being investigated. Karatzas et al. (2008) showedthat disinfectants commonly used on swine farms se-lect for MDR Salmonella. Ricci and Piddock (2009)showed that Ciprofloxacin can select for genes par-tially responsible for MDR Salmonella.
In summary, we conclude that currently availabledata do not support the hypothesis that discontinuinguse of tetracyclines in food animals would reduce riskof either MRSA infections or MDR Salmonella infec-tions in human patients. That Salmonella prevalenceis reduced approximately four-fold in swine rearedwith antibiotics in the United States, as comparedto ABF swine (Gebreyes, 2008), also suggests po-tential benefits from continued use. Since we findno evidence of increased treatment failures in hu-man patients due to co-selection of MRSA or MDRSalmonella from tetracycline use in food animals, asan alternative hypothesis, we turn next to potentialhazards caused directly by tetracycline resistance inhuman medicine.
8.3. Assessing Potential Hazards from TetracyclineResistance in Human Medicine
Tetracyclines have been used in human medicineto treat respiratory tract infections such as pneu-monia; infections of skin, genital, and urinary sys-tems; and infections that cause stomach ulcers(Helicobacter pylori). Doxycycline is the first-linetreatment for Rickettsia diseases (spotted fever, ty-phus, and scrub typhus), including Rocky Mountainspotted fever in the United States; chloramphenicolis a second-line treatment (www.merck.com/mmpe/sec14/ch177/ch177a.html). Doxycycline is an alter-native treatment for Lyme disease and, along withciprofloxacin, is recommended by the CDC as aprophylaxis against inhalation-derived anthrax infec-tions (Bell et al., 2002). FDA’s Guidance 152 doc-ument (FDA, 2003) lists tetracyclines as “HighlyImportant” in human medicine (on a scale of “Criti-cally Important,” “Highly Important,” and “Impor-tant”); this designation includes the specific drugstetracycline, chlortetracycline, demeclocycline, doxy-cycline, and minocycline. The rationale checkedin the justification column is “Sole/limited therapy
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 443
or essential therapy for serious diseases,” and thecomments column (examples) specifies “Rickettsialdisease: Anthrax therapy/prophylaxis.” However, inreality, many other antibiotics, can be used to treatanthrax infections (AHFS, 2008; Zeichner, 1998; Bellet al., 2002). Tetracycline is sometimes mentionedas a treatment option for “traveler’s diarrhea,” i.e.,E. coli enteritis (Klein and Cunha, 1995; Sack et al.,1978), but most such cases are self-limiting and re-solve within 1–3 days without medication (NIH,2008).
In addition to killing or inhibiting growth ofbacteria, tetracyclines are effective for prophylaxisand treatment of malaria due to Plasmodium falci-parum, including that due to mefloquine-resistant P.falciparum (Roberts, 2003). They also have usefulanti-inflammation, immunosuppression, dental, andwound-healing effects that are useful for treatingnoninfectious conditions, such as rosacea, and thatenhance their effectiveness in treating acne. Tetra-cyclines are one of the cheapest classes of antibi-otics available today, making them attractive foruse in developing countries with limited health carebudgets.
8.3.1. Tetracycline Prescription Ratesin Europe and the United States
In the United States, tetracycline drugs are pre-scribed for outpatient use at a higher rate than anyother antibacterial, except for penicillins, at 4.63defined daily doses per 1,000 population per day(DDDs) in 2004 (Goossens et al., 2007). This repre-sented 18.6% of total DDDs for all outpatient an-tibiotics in the United States. In Europe, the aver-age tetracycline outpatient prescription rate for 2004is 2.37 DDDs, representing 12.42% of all antibi-otics (ibid). A significant portion of this tetracyclineuse goes to treating acne and rosacea, as these of-ten require prolonged treatment periods. Acne treat-ment guidelines indicate taking oral antibiotics forbetween 3 and 6 months (Helms et al., 2006). Us-ing prescription survey data from Stern (2000), andassuming that each prescription provides an averageof 90 days of tetracycline use, yields an estimate of157,680,000 doses prescribed annually. Based on theU.S. population at the time of the survey, this equatesto 1.58 DDDs for tetracycline use in treating acne,roughly one-third of all human tetracycline use in theUnited States.
8.3.2. Survey of Tetracycline Resistance Hazardsfor Human Medical Uses of Tetracyclines
Table I summarizes current human uses of tetra-cyclines worldwide, as both primary and alternativeantibiotic selections (Roberts, 2003). It notes the U.S.incidence of each disease or condition being treated,as well as any resistance to tetracyclines by the under-lying organism. It is striking that most of these condi-tions are not relevant as potential hazards (i.e., sourcesof human treatment failure risk) from animal tetracy-cline use in the United States.
This is because the underlying conditions arerare or nonexistent in the United States; or be-cause tetracycline resistance remains rare or nonex-istent; or because the condition is noninfectious (e.g.rosacea). It is unlikely that continued tetracyclineuse would suddenly increase the tetracycline resis-tance rates for these or other organisms in the UnitedStates, as tetracyclines have already been extensivelyused for decades. Table I lists specific reasons forexcluding various conditions from further consider-ation. Estimated average annual cases, when orig-inally provided as rates, are computed with an as-sumed U.S. population of 305M (76% of which isadult); when cases are originally provided only inthe form of ranges, the midpoint of each range isassumed to be the average. A few rows indicateconditions for which tetracycline resistance cannoteasily be classified as a negligible hazard in humanmedicine. These are analyzed further below.
Table I identifies only three organisms/condi-tions that could potentially create a nonzero risk oftetracycline resistance-related treatment failures atpresent in the United States. Among 28 identified or-ganisms/conditions for which tetracycline is a treat-ment option, only the following three were identi-fied as posing potential resistance hazards to effectivetherapy:
• Periodontal disease (gingivitis) a common con-dition for which tetracyclines are sometimesused as part of treatment, and for which resis-tance is both observable before treatment andinducible during treatment;
• Acne vulgaris (acne), which is very common,and for which tetracycline treatments inducesignificant resistance in vivo in the causativeagent (P. acnes), as well as in the oral and fecalflora of patients and their family members;
• Possibly, Mycobacterium fortuitum, which isrelatively rare in the United States, but in
444 Cox and PopkenT
able
I.In
itia
lScr
eeni
ngof
Pot
enti
alH
uman
Hea
lth
Haz
ards
Due
toT
etra
cycl
ine
Res
ista
nce
Org
anis
m/C
ondi
tion
Com
mon
Nam
eU
.S.C
ases
per
Yea
rT
reat
men
tsR
esis
tanc
eto
Tet
racy
clin
e
Yer
sini
ape
stis
Pro
phyl
axis
/tre
atm
ent
Pla
gue
1to
40ca
ses
(avg
=13
case
s)by
wes
tern
stat
es,1
971–
1995
.htt
p://
ww
w.c
dc.g
ov/n
cido
d/dv
bid/
plag
ue/f
acts
.htm
Stre
ptom
ycin
isth
ean
tibi
otic
ofch
oice
.G
enta
mic
inis
used
ifst
rept
omyc
inis
nota
vaila
ble.
Tet
racy
clin
esan
dch
lora
mph
enic
olar
eal
soef
fect
ive.
ww
w.c
dc.g
ov/n
cido
d/dv
bid/
plag
ue/
qa.h
tm
Ext
rem
ely
rare
:2do
cum
ente
dca
ses
wor
ldw
ide,
none
inth
eU
nite
dSt
ates
.ht
tp://
aac.
asm
.org
/cgi
/con
tent
/ful
l/50/
10/3
233#
R28
none
Bac
illus
anth
raci
sP
rim
ary
trea
tmen
t/al
tern
ativ
epr
ophy
laxi
s
Ant
hrax
∼1–
2(a
ssum
eav
g=
1.5)
case
sof
cuta
neou
sdi
seas
e(2
2ca
ses
in20
01/2
002
terr
orat
tack
).ht
tp://
ww
w.c
dc.g
ov/n
czve
d/df
bmd/
dise
ase
listi
ng/a
nthr
axti
.htm
l
Cip
roflo
xaci
n,do
xycy
clin
e,am
pici
llin,
peni
cilli
n,cl
inda
myc
in,
clar
ithr
omyc
in,i
mip
enem
/m
erop
enem
,van
com
ycin
,rif
ampi
cin,
and
chlo
ram
phen
icol
.htt
p://w
ww
.pu
bmed
cent
ral.n
ih.g
ov/a
rtic
lere
nder
.fc
gi?a
rtid
=176
9905
Non
eob
serv
edin
2001
/200
2ca
ses
orin
65hi
stor
ical
isol
ates
held
byC
DC
(Bel
let
al.,
2002
),bu
tsom
ewha
tind
ucib
lein
vitr
o.ht
tp://
jac.
oxfo
rdjo
urna
ls.
org/
cgi/c
onte
nt/f
ull/5
4/2/
424.
Thu
sca
utio
nis
inor
der
inth
eca
seof
long
-ter
mtr
eatm
ent.
Fra
ncis
ella
tula
rens
isP
rim
ary
prop
hyla
xis/
alte
rnat
ive
trea
tmen
t
Rab
bitf
ever
100–
200
(ass
ume
avg
=15
0).h
ttp:
//w
ww
.cdc
.gov
/nci
dod/
EID
/vol
11no
12/0
5-07
28.h
tm
Dox
ycyl
ine,
Cip
roflo
xaci
n.ht
tp://
ww
w.
bt.c
dc.g
ov/a
gent
/tul
arem
ia/t
ular
emia
-bi
olog
ical
-wea
pon-
abst
ract
.asp
#4
Susc
epti
ble
inN
orth
Am
eric
ansa
mpl
es.
http
://aa
c.as
m.o
rg/c
gi/c
onte
nt/f
ull/5
2/6/
2276
Bor
relia
burg
dorf
eri
Pri
mar
ytr
eatm
ent
Lym
edi
seas
e20
,000
.htt
p://w
ww
.cdc
.gov
/mm
wr/
prev
iew
/mm
wrh
tml/m
m56
23a1
.ht
m
Dox
ycyc
line,
amox
icill
in,o
rce
furo
xim
eax
etil.
http
://w
ww
.cdc
.gov
/nci
dod/
dvbi
d/L
YM
E/ld
hum
andi
seas
etr
eatm
ent.h
tm
Susc
epti
ble
(Ger
man
and
Fre
nch
rese
arch
).ht
tp://
ww
w.s
prin
gerl
ink.
com
/con
tent
/p72
0010
6048
7246
4/,
http
://ca
t.ini
st.fr
/?aM
odel
e=af
fiche
N&
cpsi
dt=1
3869
51B
arto
nella
Qui
ntan
aP
rim
ary
trea
tmen
tT
renc
hfe
ver
“Afe
wca
ses”
asso
ciat
edw
ith
imm
unoc
ompr
omiz
edpa
tien
ts,
espe
cial
lyH
IVpa
tien
tsan
dho
mel
ess
alco
holic
s(a
ssum
eav
g=
10).
http
://w
ww
.cdc
.gov
/nc
idod
/eid
/vol
2no2
/jack
son.
htm
Imm
unoc
ompr
omis
ed:u
nest
ablis
hed
Imm
unoc
ompe
tent
:ery
thro
myc
in,
azit
hrom
ycin
,dox
ycyc
line,
orte
trac
yclin
e.ht
tp://
ww
w.c
dc.g
ov/
ncid
od/e
id/v
ol2n
o2/ja
ckso
n.ht
m
Susc
epti
ble.
http
://aa
c.as
m.o
rg/c
gi/
cont
ent/
full/
48/6
/192
1
Bar
tone
llahe
nsel
aeP
rim
ary
trea
tmen
tC
at-s
crat
chdi
seas
e1
per
10,0
00pe
rson
s.ht
tp://
ww
w.
emed
icin
e.co
m/M
ED
/top
ic21
2.ht
m(a
vg=
30,5
00)
CSD
isty
pica
llya
beni
gn,s
elf-
limit
edill
ness
last
ing
6–12
wee
ksin
the
abse
nce
ofan
tibi
otic
ther
apy
(90%
).E
ryth
rom
ycin
and
rifa
mpi
nar
eth
ean
tibi
otic
sof
choi
cein
hum
ans.
Susc
epti
ble.
http
://aa
c.as
m.o
rg/c
gi/
cont
ent/
full/
48/6
/192
1
Bru
cella
spec
ies
Pri
mar
ytr
eatm
ent
Bru
cello
sis
100–
200
(ass
ume
avg
=15
0).h
ttp:
//w
ww
.cdc
.gov
/nci
dod/
dbm
d/di
seas
einf
o/B
ruce
llosi
sg.
htm
#how
com
mon
Dox
ycyc
line
wit
hge
ntam
icin
.htt
p://
med
icin
e.pl
osjo
urna
ls.o
rg/p
erls
erv/
?req
uest
=get
-doc
umen
t&do
i=10
.137
1
Not
cons
ider
eda
heal
this
sue.
http
://m
edic
ine.
plos
jour
nals
.org
/per
lser
v/?r
eque
st=g
etdo
cum
ent&
doi=
10.1
371/
jour
nal.p
med
.004
0317
&ct
=1R
icke
ttsia
spec
ies
Pri
mar
ytr
eatm
ent
Spot
ted
feve
r(R
ocky
Mou
ntai
nSp
otte
dF
ever
)
250–
1,20
0ca
ses
ofre
port
edan
nual
ly(a
ssum
eav
g=
725)
.w
ww
.cdc
.gov
/tic
ks/d
isea
ses/
rock
ym
ount
ain
spot
ted
feve
r/st
atis
tics
.htm
l
Dox
ycyc
line.
http
://cm
r.as
m.o
rg/c
gi/
cont
ent/
full/
18/4
/719
Non
eob
serv
edto
date
(Rob
erts
,200
3);
see
also
Lor
ian
(200
5,p.
275) (C
ontin
ued)
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 445T
able
I.(C
ontin
ued)
Org
anis
m/C
ondi
tion
Com
mon
Nam
eU
.S.C
ases
per
Yea
rT
reat
men
tsR
esis
tanc
eto
Tet
racy
clin
e
Myc
opla
sma
pneu
mon
iaP
rim
ary
trea
tmen
t
Chl
amyd
iapn
eum
onia
Pri
mar
ytr
eatm
ent
Aty
pica
lpne
umon
ia
Aty
pica
lpne
umon
ia
2m
illio
nca
ses
and
100,
000
pneu
mon
ia-r
elat
edho
spit
aliz
atio
nsoc
cur
inth
eU
nite
dSt
ates
.htt
p://w
ww
.cdc
.go
v/nc
idod
/dbm
d/di
seas
einf
o/m
ycop
lasm
apne
umt.h
tmA
typi
calp
neum
onia
(we
incl
ude
2of
the
3m
ajor
vari
ants
)m
ayac
coun
tfor
upto
40%
ofco
mm
unit
y-ac
quir
edpn
eum
onia
case
s.ht
tp://
ww
w.
med
scap
e.co
m/v
iew
arti
cle/
4232
25(a
ssum
eav
g=
800,
000)
Seve
rala
ntim
icro
bial
sar
eef
fect
ive
inre
duci
ngth
ele
ngth
ofill
ness
due
tom
ycop
lasm
alpn
eum
onia
,inc
ludi
nger
ythr
omyc
in,a
zith
rom
ycin
,cl
arit
hrom
ycin
,dox
ycyc
line,
and
quin
olon
es
Dox
ycyl
cine
isfir
stch
oice
,mac
rolid
esar
ean
alte
rnat
e,bu
tmay
bele
ssef
fect
ive.
ww
w.e
med
icin
e.co
m/m
ed/
topi
c341
.htm
To
date
,no
clin
ical
isol
ates
resi
stan
tto
tetr
acyc
lines
have
been
desc
ribe
din
the
liter
atur
e,bu
tsom
eha
vebe
ense
lect
edin
vitr
o.ht
tp://
ww
w.in
gent
acon
nect
.com
/con
tent
/ben
/cdt
id/
2005
/000
0000
5/00
0000
03/a
rt00
006
Not
obse
rved
inC
hlam
ydia
spec
ies
(Rob
erts
,200
3).h
ttp:
//ww
w.c
hlam
ydi
ae.c
om/r
estr
icte
d/do
cs/la
btes
ts/
trea
tan
tibi
otic
res.
asp
Chl
amyd
iaps
ittac
iP
rim
ary
trea
tmen
tO
rnit
hosi
s(f
rom
bird
s)∼8
1.ht
tp://
ww
w.e
med
icin
e.co
m/
med
/top
ic34
1.ht
mT
etra
cycl
ine
ordo
xycy
clin
eis
the
trea
tmen
tofc
hoic
e,er
ythr
omyc
inis
the
alte
rnat
ive
trea
tmen
t.w
ww
.em
edi
cine
.com
/med
/top
ic34
1.ht
m
Not
obse
rved
inC
hlam
ydia
spec
ies
(Rob
erts
,200
3).h
ttp:
//ww
w.c
hlam
ydi
ae.c
om/r
estr
icte
d/do
cs/la
btes
ts/
trea
tan
tibi
otic
res.
asp
Chl
amyd
iatr
acho
mat
isP
rim
ary
trea
tmen
tIn
fant
pneu
mon
iaIn
fant
s:∼
12,0
00(c
ontr
aind
icat
ed).
Adu
ltca
ses
have
been
repo
rted
inim
mun
ocom
prom
ised
host
s.w
ww
.em
edic
ine.
com
/med
/top
ic34
1.ht
m(a
ssum
eav
g=
10)
Tre
atw
ith
eryt
hrom
ycin
.Dox
ycyc
line
isco
ntra
indi
cate
din
child
ren
youn
ger
than
9ye
ars.
http
://w
ww
.em
edic
ine.
com
/med
/top
ic34
1.ht
m
Not
obse
rved
inC
hlam
ydia
spec
ies
(Rob
erts
,200
3).h
ttp:
//ww
w.c
hlam
ydi
ae.c
om/r
estr
icte
d/do
cs/la
btes
ts/
trea
tan
tibi
otic
res.
asp
Mefl
oqui
ne-r
esis
tant
Pla
smod
ium
falc
ipar
umP
rim
ary
trea
tmen
tal
tern
ativ
epr
ophy
laxi
s
Avi
rule
ntty
peof
mal
aria
Non
e(a
ttem
pera
ture
sbe
low
68◦ F
,Pla
smod
ium
falc
ipar
umca
nnot
com
plet
eit
sgr
owth
cycl
ein
Ano
phel
esm
osqu
ito,
soca
nnot
betr
ansm
itte
d).w
ww
.cd
c.go
v/M
alar
ia/d
istr
ibut
ion
epi/d
istr
ibut
ion.
htm
(ass
ume
avg
∼0)
Mefl
oqui
ne,q
uini
nein
com
bina
tion
wit
hdo
xycy
clin
e,or
Fan
sida
rR©
.ht
tp://
ww
w.tu
lane
.edu
/∼w
iser
/pro
tozo
olog
y/no
tes/
mal
aria
.htm
l
Susc
epti
ble.
http
://w
ww
.spr
inge
rlin
k.co
m/c
onte
nt/9
0v37
9gry
1l2m
hlh/
,ht
tp://
ww
w.a
jtm
h.or
g/cg
i/con
tent
/ab
stra
ct/4
7/1/
108
Bal
antid
ium
coli
Pri
mar
ytr
eatm
ent
Cili
ated
prot
ozoa
nB
alan
tidi
asis
isfo
und
wor
ldw
ide
and
has
anov
eral
lest
imat
edpr
eval
ence
of1%
buti
sra
rein
the
Uni
ted
Stat
es.M
ostc
ases
ofba
lant
idia
sis
inim
mun
ocom
pete
ntin
divi
dual
sar
eas
ympt
omat
ic.w
ww
.em
edi
cine
.com
/ME
D/t
opic
203.
htm
(ass
ume
avg
∼0)
Tet
racy
clin
eis
the
trea
tmen
tofc
hoic
e,w
ith
met
roni
dazo
lebe
ing
the
prim
ary
alte
rnat
ive.
Iodo
quin
ol,p
urom
ycin
,an
dni
tazo
xani
dear
eal
soef
fect
ive
agai
nstb
alan
tidi
asis
.ww
w.
emed
icin
e.co
m/M
ED
/top
ic20
3.ht
m
No
tetr
acyc
line-
resi
stan
tpro
tozo
ans
have
yetb
een
desc
ribe
d(R
ober
ts,2
003)
(Con
tinue
d)
446 Cox and Popken
Tab
leI.
(Con
tinue
d)
Org
anis
m/C
ondi
tion
Com
mon
Nam
eU
.S.C
ases
per
Yea
rT
reat
men
tsR
esis
tanc
eto
Tet
racy
clin
e
Per
iodo
ntal
dise
ase
Pri
mar
ytr
eatm
ent
Gin
givi
tis
44%
to57
%of
adul
tsha
vem
oder
ate
peri
odon
titi
s,an
dab
out7
%to
15%
have
adva
nced
peri
odon
titi
s(B
eck
and
Off
enba
cher
,200
3)(a
ssum
eav
g∼
27M
)
Mec
hani
calp
roce
dure
s,or
alri
nses
,and
som
etim
ean
tibi
otic
sin
clud
ing
tetr
acyc
lines
(Am
eric
anA
cade
my
ofP
erio
dont
olog
y,20
01)
Vill
idie
uet
al.(
2003
)fo
und
11%
ofor
alm
icro
bial
flora
inhe
alth
yad
ults
cont
ains
tetr
acyc
line
resi
stan
cege
nes.
Ols
vik
and
Ten
over
(199
3)fo
und
upto
75%
ofba
cter
iain
subg
ingi
valfl
ora
may
cont
ain
tetr
acyc
line
resi
stan
cege
nes
afte
rlo
ng-t
erm
trea
tmen
t
Acn
evu
lgar
is(P
.acn
es)
Pri
mar
ytr
eatm
ent
Acn
e5.
1m
illio
nan
tibi
otic
pres
crip
tion
for
acne
annu
ally
betw
een
1996
and
1998
(Ste
rn,2
000)
Com
bina
tion
trea
tmen
tsin
clud
ing
tret
inoi
n,be
nzoy
lper
oxid
e,to
pica
lan
tibi
otic
s(e
ryth
rom
ycin
,cl
inda
myc
in),
oral
anti
biot
ics
(tet
racy
clin
ce,m
inoc
yclin
e,er
ythr
omyc
in,d
oxyc
yclin
e,tr
imet
hopr
im-s
ulfa
met
hoxa
zole
),an
dis
otre
tino
in(H
elm
set
al.,
2006
)
Indu
ced
bybo
thto
pica
l(R
ober
ts,2
003)
and
oral
trea
tmen
ts.R
esis
tanc
efo
und
inP
.Acn
es(S
wan
son,
2003
)an
dbo
thor
al(R
ober
ts,2
003)
and
inte
stin
alflo
ra(V
alto
nen
etal
.,19
76).
Pat
ient
san
dfa
mily
mem
bers
(Ada
ms
etal
.,19
85)
affe
cted
.Res
ista
ntva
riet
ies
appe
artr
ansm
issi
ble
betw
een
acne
-pr
one
indi
vidu
als
(Ros
set
al.,
2003
)
Blis
teri
ngsk
indi
seas
esP
rim
ary
trea
tmen
tN
A(n
onin
fect
ious
)
Ros
acea
Pri
mar
ytr
eatm
ent
1.1
mill
ion
outp
atie
ntvi
sits
per
year
.ww
w.n
cbi.n
lm.n
ih.g
ov/
pubm
ed/1
1534
91
20.4
9%of
pati
ents
use
tetr
acyc
lines
(Rom
anow
icz
etal
.,20
08)
NA
(non
infe
ctio
us)
Vib
rio
chol
era
Alt
erna
tive
trea
tmen
tC
hole
ra“V
irtu
ally
elim
inat
ed.”
http
://w
ww
.cdc
.gov
/ncz
ved/
dfbm
d/di
seas
elis
ting
/cho
lera
gi.h
tml(
assu
me
avg
∼0)
Cho
lera
can
besi
mpl
yan
dsu
cces
sful
lytr
eate
dby
imm
edia
tere
plac
emen
tof
the
fluid
and
salt
slo
stth
roug
hdi
arrh
ea.A
ntib
ioti
cssh
orte
nth
eco
urse
and
dim
inis
hth
ese
veri
tyof
the
illne
ss,b
utth
eyar
eno
tas
impo
rtan
tas
rehy
drat
ion
Rar
e,bu
tem
ergi
ngin
som
epa
rts
ofth
ew
orld
.htt
p://w
ww
.cdc
.gov
/Nci
dod/
EID
/vol
8no3
/01-
0258
.htm
Lep
tosp
ira
spec
ies
Alt
erna
tive
trea
tmen
tW
eild
isea
se0.
02–0
.04
case
spe
r10
0,00
0pe
rson
sin
1985
–199
4.ht
tp://
ww
w.e
med
icin
e.co
m/p
ed/
byna
me/
Lep
tosp
iros
is.h
tm(a
ssum
eav
g=
92)
Pen
icill
in,d
oxyc
yclin
e,am
pici
llin,
amox
icill
inN
ore
port
sof
resi
stan
cefo
rco
mm
onan
tibi
otic
s.w
ww
.who
.int/
zoon
oses
/di
seas
es/L
epto
spir
osis
surv
eilla
nce.
Tre
pone
ma
palli
dum
Alt
erna
tive
trea
tmen
tSy
phili
s36
,000
case
sof
syph
ilis
in20
06,
incl
udin
g9,
756
case
sof
prim
ary
and
seco
ndar
y(P
&S)
syph
ilis.
ww
w.c
dc.g
ov/s
td/s
yphi
lis/S
TD
Fac
t-Sy
phili
s.ht
m
Pen
icill
in,o
ther
anti
biot
ics
ifpa
tien
tis
alle
rgic
(dox
ycyc
line,
tetr
acyc
line,
and
eryt
hrom
ycin
)
Non
ere
port
ed.h
ttp:
//ww
w.c
dc.g
ov/s
td/
trea
tmen
t
(Con
tinue
d)
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 447T
able
I.(C
ontin
ued)
Org
anis
m/C
ondi
tion
Com
mon
Nam
eU
.S.C
ases
per
Yea
rT
reat
men
tsR
esis
tanc
eto
Tet
racy
clin
e
Hel
icob
acte
rpy
lori
Alt
erna
tive
trea
tmen
t“S
tom
ach
ulce
r”E
ach
year
ther
ear
e50
0,00
0to
850,
000
new
case
sof
pept
icul
cer
dise
ase
and
mor
eth
an1
mill
ion
ulce
r-re
late
dho
spit
aliz
atio
ns(a
ssum
eav
g=
1M)
Tet
racy
clin
esar
eno
tfirs
t-lin
etr
eatm
entf
orH
.pyl
ori.
Mul
tidr
ugth
erap
ies
are
mor
eef
fect
ive
(Mir
bagh
erie
tal.,
2006
).T
etra
cycl
ine
isa
sugg
este
dco
mpo
nent
inso
me
com
bina
tion
ther
apie
s.w
ww
.hel
ico.
com
/tre
atth
erap
y.ht
ml
Not
foun
din
the
Uni
ted
Stat
es.h
ttp:
//w
ww
.pub
med
cent
ral.n
ih.g
ov/a
rtic
lere
nder
.fcgi
?art
id=1
3277
8
Myc
obac
teri
umm
arin
umA
lter
nati
vetr
eatm
ent
Fis
hta
nkgr
anul
oma
Est
imat
edan
nual
inci
denc
eis
0.27
case
spe
r10
0,00
0ad
ultp
atie
nts
(ass
ume
avg
=82
4).w
ww
.em
edi
cine
.com
/med
/top
ic15
38.h
tm
Thi
sor
gani
smis
sens
itiv
eto
rifa
mpi
n,et
ham
buto
l,te
trac
yclin
es,
trim
etho
prim
-sul
fam
etho
xazo
le(T
MP
-SM
X),
clar
ithr
omyc
in,a
ndle
voflo
xaci
n
Not
sign
ifica
nt.h
ttp:
//ww
w.n
cbi.n
lm.n
ih.
gov/
pubm
ed/1
2153
378?
dopt
=Abs
trac
t
Myc
obac
teri
umfo
rtui
tum
Alt
erna
tive
trea
tmen
tF
ound
inna
tura
lan
dpr
oces
sed
wat
erso
urce
s,as
wel
las
inse
wag
ean
ddi
rt
Fro
m19
93–1
996,
4.65
–5.9
9ca
ses
per
mill
ion
pers
ons
wer
ere
port
edto
the
CD
C(F
ritz
and
Woe
ltje
,200
7)(a
ssum
eav
g∼
1,62
3)
Pro
long
edan
tibi
otic
ther
apy
isge
nera
llyre
quir
edfo
rM
fort
uitu
min
fect
ion.
Man
yan
tibi
otic
sus
ed.
Bec
ause
doxy
cycl
ine
has
acti
vity
agai
nsto
nly
appr
oxim
atel
yon
e-th
ird
ofis
olat
es,g
ener
ally
notu
sed
aspa
rtof
init
iale
mpi
ric
regi
men
Rat
esof
resi
stan
ceup
to50
%ha
vebe
enre
port
ed.h
ttp:
//pro
d.ho
pkin
s-ab
xgui
de.o
rg/p
atho
gens
/bac
teri
a/m
yco
bact
eria
/myc
obac
teri
umfo
rtui
tum
.ht
ml?
cont
entI
nsta
nceI
d=25
5909
Bur
khol
deri
asp
ecie
sA
lter
nati
vetr
eatm
ent
Gla
nder
san
dm
elio
idos
is(b
iow
eapo
n?)
Afe
wis
olat
edca
ses
per
year
.ht
tp://
ww
w.c
dc.g
ov/n
czve
d/df
bmd/
dise
ase
listi
ng/m
elio
ido
sis
gi.h
tml(
assu
me
avg
=3)
Bur
khol
deri
aps
uedo
mal
lei,
the
orga
nism
that
caus
esm
elio
idos
is,i
sus
ually
sens
itiv
eto
imip
enem
,pe
nici
llin,
doxy
cycl
ine,
amox
icill
in-c
lavu
lani
cac
id,a
zloc
illin
,ce
ftaz
idim
e,ti
carc
illin
-vul
anic
acid
,ce
ftri
axon
e,an
daz
treo
nam
Res
ista
nce
gene
spo
ssib
lyw
ides
prea
d.ht
tp://
ww
w.js
tage
.jst.g
o.jp
/art
icle
/js
me2
/22/
1/22
44/
arti
cle
Don
ovan
iagr
anul
omat
isA
lter
nati
vetr
eatm
ent
Don
ovan
osis
(ST
D)
Few
erth
an10
0ca
ses
are
repo
rted
annu
ally
,man
yof
whi
char
eth
ough
tto
bedu
eto
fore
ign
trav
el.h
ttp:
//ww
w.e
med
icin
e.co
m/d
erm
/top
ic17
2.ht
m(a
ssum
eav
g=
100)
The
reco
mm
ende
dan
tibi
otic
isei
ther
trim
etho
prim
/sul
fam
etho
xazo
le1
ordo
xycy
clin
e.A
lter
nati
ves
incl
ude
cipr
oflox
acin
,ery
thro
myc
in,o
raz
ithr
omyc
in
Rar
e.ht
tp://
ww
w.p
ubm
edce
ntra
l.nih
.gov
/ar
ticl
eren
der.
fcgi
?art
id=1
1947
66
Epi
didy
mit
isA
lter
nati
vetr
eatm
ent
Scro
tali
nfec
tion
,va
riou
sca
uses
Inci
denc
eis
less
than
1ca
sein
1,00
0m
ales
per
year
Var
iety
ofan
tibi
otic
s.ht
tp://
ww
w.
emed
icin
e.co
m/e
mer
g/to
pic1
66.h
tmIn
dete
rmin
ate
(man
yin
fect
ious
and
noni
nfec
tiou
sca
uses
)
Pro
stat
itis
Alt
erna
tive
trea
tmen
tP
rost
ate
infla
mm
atio
n,va
riou
sca
uses
2m
illio
nou
tpat
ient
visi
tspe
rye
ar.
http
://w
ww
.em
edic
ine.
com
/em
erg/
topi
c488
.htm
Var
ious
anti
biot
ics
are
usua
llyfir
stch
oice
oftr
eatm
ent.
4–8
wee
ksof
pros
tate
-pen
etra
ting
anti
biot
ics,
such
fluor
oqui
nolo
neor
trim
etho
prim
-su
lfam
etho
xazo
le.w
ww
.may
oclin
ic.
com
/hea
lth/
pros
tati
tis/
DS0
0341
/dse
ctio
n=tr
eatm
ents
-and
-dru
gsw
ww
.em
edic
ine.
com
/em
erg/
topi
c488
.htm
Inde
term
inat
e(m
any
infe
ctio
usan
dno
ninf
ecti
ous
caus
es)
448 Cox and Popken
which high resistance rates to tetracyclines (analternative treatment) have been observed.
The potential hazard from each of these three condi-tions is assessed in more detail next.
8.3.3. Tetracycline Resistance Hazards in TreatingPeriodontal Disease
Systemic antibiotics, including tetracyclines, areused to treat a subset of cases of severe gin-givitis or chronic periodontal disease. Lundergan(2003) describes these cases as “aggressive” or “re-fractory” periodontitis, and notes that tetracyclinehydrochloride, doxycycline, metronidazole, clin-damycin, ciprofloxacin, amoxicillin/clavulanic acid,mitronidazole in combination with amoxicillin, andmetronidazole with ciprofloxacin have all been pre-viously used. Greenstein (2000) provides the samelist of potential systemic antibiotics. Both authorsnote the need for species identification and prior sus-ceptibility testing before selecting an antibiotic tomaximize effectiveness of treatment. The wide va-riety of bacteria (more than 500) identified withinperiodontal pockets, even when narrowed to onlythose specifically associated with periodontal dis-ease, are still quite diverse. For example, for A.actinomycetemcomitans–associated periodontitis, re-searchers recommend that clinicians use amoxicillinwith clavulanic acid and metronidazole. If the pa-tient is allergic to penicillin, ciprofloxacin couldbe substituted for amoxicillin with clavulanic acid.Ciprofloxacin is effective against enteric rods, pseu-domonads, and staphylococci. Another antibioticthat is effective and specific for anaerobes is clin-damycin (Greenstein, 2000). At the same time,Greenstein concludes that systemic antibiotic ther-apy is usually not needed and should be avoided inroutine treatment.
Although tetracycline resistance in bacteria as-sociated with periodontal disease may be inducedby treatment, it is unclear to what extent such resis-tance translates into treatment failures. Kleinfelderet al. (1999) found in their study of subgingivialplaque microorganisms that tetracycline resistancerates among the major species identified ranged from3 to 29%. However, the MICs for every sample werewell below the concentration achievable in gingivalcrevicular fluid, implying that treatment could stillsucceed. Rodrigues et al. (2004) found that locallyor systemically administered tetracycline resulted ina temporary selection of subgingival species intrinsi-
cally resistant to tetracycline. However, both thera-pies were effective in reducing their prevalence overtime.
Part of the apparent success of tetracyclines intreating periodontal disease, even in the presence oftetracycline resistant oral bacteria, may be explainedby nonbactericidal healing effects. Tetracyclines re-duce inflammation and help block collagenase, anenzyme that destroys connective bone tissue. Theycan also promote the regeneration of lost periodon-tal tissue. These characteristics enable subantimicro-bial doses of doxycycline to be part of an effectivetreatment regimen (Greenstein, 2000).
Based on this information, we conclude thatthere is no identified risk of periodontal disease treat-ment failures due to tetracycline resistance, regard-less of its source. Even if there is an undetected, butnonzero, risk of compromised treatment, it wouldsurely be dominated by treatment-related selectionof resistant bacteria; however, even in the presenceof such resistance, tetracycline therapy appears to beeffective (Rodrigues et al., 2004).
8.3.4. Tetracycline Resistance Hazards in TreatingMycobacterium Fortuitum
Mycobacterium fortuitum is a nontuberculousmycobacterium (NTM) found in natural and pro-cessed water sources, as well as in sewage anddirt. Distribution is probably worldwide (Fritz andWoeltje, 2007). It has various clinical expressions, in-cluding local cutaneous disease, osteomyelitis, jointinfections, and ocular disease, and rarely, lung dis-ease and lymphadenitis. Treatment requires pro-longed (6 months) antibiotic therapy, intravenously,at least initially, for serious cases.
Mycobacterium fortuitum is not a reportable dis-ease in the United States, so its incidence is diffi-cult to determine. Fritz and Woeltje (2007) estimatethat 4.65–5.99 cases occurred per million populationduring 1993–1996, equating to a current level of ap-proximately 450 cases per year. Roberts (2003) re-ported tetracyclines as an alternative treatment forM fortuitum infections. However, current treatmentrecommendations describe Amikacin, an aminogly-coside, as the preferred treatment for M fortuituminfections; almost all isolates are susceptible. Bothcefoxitin and imipenem have been used successfully,but susceptibility is variable. Ciprofloxacin and lev-ofloxacin have both been used successfully (Fritz andWoeltje, 2007). Antibiotic resistance in M. fortui-tum is highly variable by subspecies. Resistance to
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 449
doxycycline ranged from 26% to 63% in a study bySwenson et al. (1985). Therefore, doxycyline is notrecommended as a first-line treatment in the absenceof susceptibility testing.
Given this information, it is unlikely that thereare any tetracycline (doxycycline) resistance-relatedtreatment failures of Mycobacterium fortuitum infec-tions in the United States. Doxycycline is not usedunless the infection is known to be susceptible, andseveral better alternatives exist. If there are no tetra-cycline resistance-related treatment failures, then therisk of such treatment failures caused by animal usesof tetracyclines would be zero.
8.3.5. Tetracycline Resistance Hazardsin Treating Acne
The Merck Manual (Merck, 2008) recommendsoral antibiotics, in conjunction with topical antibi-otics, for patients with moderate acne. For these pa-tients, “tetracycline is usually a good first choice.”Doxycycline and erythromycin are recommended assecond-line agents. Nontetracyline treatments arerecommended for other acne patients, including: top-ical clindamycin and benzoyl peroxide gels for mildinflammatory acne; oral isotretinoin for severe acne;and triamcinolone for cystic acne.
8.3.6. Historical Tetracycline Usage and Resistancefor Acne in the United States and England
Tetracyclines have been used to treat millionsof cases of acne in the United States for decades.Stern (2000) showed 5.1 million prescriptions oforal antibiotics (both new and renewals) for acnealone between 1996 and 1998 in the United States—approximately 2,876,00 women and 2,220,000 men.The data provided for women’s use of oral antibioticsshow that just over a third of the prescriptions werefor tetracyclines, broken down as: 13% for minocy-cline, 14% for tetracycline, and 7.5% for doxycycline.This extensive tetracycline use in human medicineis generally considered to have caused the spreadof tetracycline-resistant strains of Propionibacteriumacnes, the bacterium that causes infectious acne (Tanand Tan, 2005).
The risk of tetracycline-resistant acne for indi-vidual patients depends on many variables (e.g., age,stress, skin type, and so forth), but has not been as-sociated with animal uses of antibiotics. (For exam-ple, a PUBMED search of “acne growth promoter”returns no hits.) One possible way to investigate
whether use of tetracycline in animals might con-tribute significantly to human tetracycline-resistantacne cases is to compare resistance rates in hu-man patients from locations with and without suchuse. (Such “ecological” comparisons may not sup-port valid causal inferences, but can nonethelessbe useful in identifying differences that can thenbe investigated more thoroughly by other methods,such as time series analysis.) Eady et al. (2003) per-formed a comprehensive evaluation of study dataon resistance of P. acnes to antibiotic treatments,and concluded that overall prevalence rates for skincolonization with tetracycline resistant strains were“as high as 30%” in the United States. This isvery consistent with a study of acne patients inthe United Kingdom, which estimated that about26.4% had tetracycline resistant P. acnes (Ross,2003). Tetracyclines have not been used as growthpromotion STAs in the United Kingdom for nearlyfour decades (since the early 1970s). Thus, it ap-pears that substantial reservoirs of resistant P. ac-nes are created and maintained by other (non-STA)uses (presumably including treatment of humanpatients).
A second way to search for possible hazards isto examine longitudinal data, e.g., how much havetreatment failures increased as animal antibiotic useand tetracycline resistance have increased? Again,detecting correlated increases does not by itself es-tablish causality, but it may indicate a pattern worthinvestigating more carefully via time series analysis,quasi-experimental designs, or other causal analysistechniques. Del Rosso and Kim (2009) describe theincrease in tetracycline resistance among acne pa-tients over the past 40 years as reducing the effec-tiveness of tetracycline as a treatment option: “Overthe past 4 decades, as the sensitivity of Propionibac-terium acnes to several oral and topical antibioticshas decreased, the efficacy of oral tetracycline anderythromycin has markedly diminished, leading to in-creased use of doxycycline, minocycline, and otheragents, such as trimethoprim/sulfamethoxazole.”On the other hand, Simonart et al. (2008) re-ported that, despite the increases in antibiotic re-sistance between 1962 and 2006, “oral tetracyclineformulations displayed no change in efficacy” for thetreatment of acne vulgaris. Thus, compounds in thetetracycline family (especially doxycycline) are stilleffective. Moreover, the increase in resistance dueto human use has been large enough to prompt de-velopment of several alternative therapies, describednext.
450 Cox and Popken
8.3.7. Alternatives to Tetracyclines and ChangingPrescription Practices
Despite their heavy historical use in treatingacne, tetracyclines are becoming regarded by manyclinicians as less desirable than other approaches fortreating most acne cases, in both the United Statesand Europe. For example, Kertzman et al. (2008)noted: “The 1999 practice guideline ‘Acne vulgaris’from the Dutch College of General Practitioners hasbeen revised. Benzoyl peroxide and local retinoidsare first choice in local treatment of acne. Whentreatment with oral antibiotics is indicated, doxycy-cline is first choice. Use of minocycline is not recom-mended in general practice. It is recommended thatboth local and oral antibiotics are always combinedwith local benzoyl peroxide or a local retinoid.” Arecent clinical trial in England concluded: “Topicalantimicrobial therapies performed at least as well asoral antibiotics in terms of clinical efficacy. Benzoylperoxide was the most cost-effective and minocyclinethe least cost-effective therapy for facial acne. Theefficacy of all three topical regimens was not com-promised by pre-existing propionibacterial resistance.Benzoyl peroxide was associated with a greater fre-quency and severity of local irritant reactions. It issuggested that the use of a combination of topicalbenzoyl peroxide and erythromycin gives less irrita-tion and better quality of life” (Ozolins et al., 2005,emphasis added).
Clindamycin is becoming increasingly popular inplace of erythromycin and tetracycline. Guay (2007)reported that “Clindamycin appears to be superiorin efficacy compared with erythromycin and tetra-cycline,” and other investigators have recently rec-ommended a combination clindamycin and benzoylperoxide topical gel as highly effective, popular withpatients, and not associated with bacterial resis-tance (e.g., Langner et al., 2008; McKeage and Keat-ing, 2008). Thus, it appears that future prescriptionsmay shift away from oral tetracyclines toward top-ical clindamycin and benzoyl peroxide gels. In theUnited States, there has also already been a signifi-cant shift away from antibiotic therapies toward top-ical retinoids and oral isotretinoin (Thevarajah et al.,2005).
In summary, although tetracycline resistance inP. acnes poses a real risk of reduced efficacy of ini-tial treatment, this risk appears to be caused pri-marily by the existence of a large, easily transmittedreservoir of tetracycline-resistant P. acnes, which, inturn, results from human use of tetracyclines to treat
acne. Tetracycline use in food animals has never beenidentified as contributing to the risk of tetracycline-resistant P. acnes. To the contrary, comparing datafrom the United States and the United Kingdom,and examining longitudinal data in the United States,suggests that human use suffices to maintain a pool oftetracycline-resistant P. acnes, and that tetracyclineSTA use has had no known incremental impact onthe efficacy of tetracyclines in treating acne vulgaris.Based on these observations, we conclude that thereis no empirical evidence suggesting that treatment ofhuman acne cases is compromised by use of tetracy-clines in food animals.
8.4. From Qualitative Hazard Assessmentto Quantitative Bounds on Risks fromTetracycline Resistance in Human Medicine
Our qualitative review of the potential tetracy-cline resistance hazards in Table I suggests that tetra-cycline STA use presents minimal or no risks to hu-man health, as each condition is either noninfectious(blistering skin diseases and rosacea), has indetermi-nate causes (epididymitis and prostatis), has no ap-parent relation to tetracycline use in food animals(acne, periodontal disease, and Mycobacterium for-tuitum), or is due to a bacterium for which resistanceto tetracyclines is not observed (all others). However,for this last category, absence of evidence of resis-tance (and of resistance-related treatment failures) isnot necessarily evidence of complete absence of re-sistance. It is only evidence that, if there is resistance,it is too rare to have been detected (so far). To ob-tain a rough upper bound on the greatest level of un-observed resistance risk that could be consistent withthe absence of observed cases, suppose that tetracy-cline resistance does exist for some of the conditionsin Table I that lack observed resistant cases. The to-tal estimated average annual number of cases fromthe 22 such conditions listed in Table I is 1,888,659.5,of which most (1,800,000) are from atypical pneumo-nia strains and Helicobacter pylori. Tetracycline re-sistance status is known only for patients for whomsamples were analyzed or tetracycline was actuallyprescribed. To get order-of-magnitude estimates ofpossible resistance risk, suppose that at least 20%of each condition leads to tetracycline prescriptionsand/or resistance screening that would reveal tetra-cycline resistance if it were present; and assume thatat least 20 years of observations are available, duringwhich tetracyclines have been used in both humanmedicine and agriculture. (In reality, both 20% and
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 451
20 years are low estimates for atypical pneumoniastrains, for which tetracycline is the usual treatment,but these values suffice to illustrate the bounding cal-culation.) Then the total number of cases in which re-sistance could have been observed, if it were present,from all 22 conditions, is about: (1,888,659.5 casesper year) ∗ (0.20 fraction observed) ∗ (20 years), orroughly 7.6 × 106 observed cases without resistance.
An approximate 95% upper confidence limit forthe true but unknown value of a proportion, basedon 0 observed cases in N binomial trials, is 3/N— theso-called rule of three for estimating risks when nooccurrences have been observed (and N > 30) (Ey-pasch et al., 1995). For N = 7.6 × 106, this formulayields an estimated 95% upper confidence limit of3/(7.5 × 106) = 4 × 10−7 for the fraction of resis-tant (but as-yet-undetected) cases. (Departures fromthe binomial model to allow a mixture of cases withdifferent resistance rates would reduce the variancearound the mean rate (Feller, 1968, p. 231). Thiscould perhaps reduce the upper confidence limit fur-ther, but we use the binomial model for simplicity,and because, among all models, it maximizes vari-ance for a given mean. Also, the estimated upperbound of 4 × 10−7 applies only as long there is nolarge change in historical rates. However, more de-tailed modeling of resistance dynamics in other con-texts suggests that a history of decades of widespreaduse without resistance emerging makes it unlikelythat a sudden jump will occur (Cox and Popken,2004).) For an individual selected at random from theU.S. population of approximately 300 million, an ap-proximate upper 95% confidence limit for the prob-ability of a tetracycline-resistant illness in a 70-yearlifetime, from any of the 22 conditions eliminatedfrom further consideration in Table I, is:
(70 years per lifetime)
∗ (1,888,659.5 cases per year
/300,000,000 population size)
∗ (potential resistant fraction of
3/[(1,888,659.5 cases per year)
∗ (0.20 fraction observed) ∗ (20 years)])
= (70/100,000,000)/(0.2 * 20) = 1.75 × 10−7.
(Note that any uncertainties about the 1,888,659.5number cancel out in this calculation, as it appearsin both the numerator and the denominator.) Themaximum-likelihood estimate, based on 0 observedresistant cases is, of course, 0. Even if such a casewere to occur, the probability that it would be pre-
scribed tetracycline (e.g., 20%, in these illustrativecalculations), times the probability that resistancewould result in treatment failure (which may be verysmall, e.g., less than 5%, based on other tetracycline-resistant infections that nonetheless respond to tetra-cycline therapy), times the probability that the resis-tant case is attributable to animal use of tetracyclines,rather than to human use (an unknown fraction, butperhaps not more than 1%, due to the much more di-rect effect of human use on human resistance), wouldfurther reduce the estimated risk to humans specif-ically from animal use of tetracycline (e.g., from ≤1.75 × 10−7 lifetime risk of a tetracycline-resistantinfection to perhaps ≤ 1.75 × 10−11 lifetime risk oftreatment failure due to a tetracycline-resistant in-fection caused by animal use of tetracycline). Evenwith the uncertainties and speculations inherent insuch calculations, it seems clear that the rough orderof magnitude of human health risks from tetracyclineuse in animals is not large. If a ban on continued usewere to reduce the risk from 1.75 × 10−11 to zero,the net effect would be to prevent only perhaps onetreatment failure every two centuries, under currentconditions.
9. TETRACYCLINE RESISTANCE INTHE ENVIRONMENT
Tetracycline-resistant bacteria are ubiquitous inthe environment, with or without animal antibioticuse (e.g., Gebreyes et al., 2006), although animalantibiotic use certainly selects for resistant strains.For example, in the United States, tetracycline- andtylosin-resistant bacteria are found in manure fromswine farms that are not using antimicrobials, as wellas in field soils where manure is applied regularly; theprevalence does not differ among farms using versusnot using antimicrobial feed additives for growth pro-motion (Chander et al., 2006). Smith et al. (2007) re-ported that resistance of E. coli to tetracycline, sul-fonamides, and streptomycin is common in broilerchickens whether or not they receive antibiotics. Sim-ilarly, a European study by Heike et al. (2006) foundthat the diversity of tetracycline resistance genes insoils augmented with pig manure was independent ofthe levels of antibiotic use at the originating farms;they concluded that there is a considerable pool ofresistance genes in soils.
Not surprisingly, tetracycline use does select fortetracycline-resistant bacteria in the environment.For example, Chee-Sanford et al. (2001) found agradient in tetracycline-resistance genes in ground
452 Cox and Popken
water, with the diversity and quantity of resistancegenes decreasing with distance from large, unlinedpig manure lagoons. Peak et al. (2007) confirmed thattetracycline usage in cattle feed lots is associated withincreased levels of tetracycline resistance genes inground water. Resistance gene abundance was highlyseasonal, 10–100 times greater in the fall than in thesummer. Engemann et al. (2008) analyzed the surviv-ability of tetracycline resistance genes in cattle feed-lot wastewater and found that resistance genes dis-appeared at a rate of 71% per day in sunlight, and51% per day in darkness. They also found significantvariation in the rates for different resistance genes.(tet(W) and tet(M) had the longest half-lives—on theorder of hours.) This study, and related work sum-marized in Graham (2005), indicates that tetracyclineresistance is developed at the point of use ratherthan in the environment. Further, resistance die-offin aquatic systems is generally very rapid, especiallyin light-exposed systems.
Taken together, these findings suggest thattetracycline use in animals selects for tetracycline-resistant bacteria in manure, soil, and water; evenwithout animal antibiotic use, however, tetracyclineresistance is widespread.
10. DISCUSSION AND CONCLUSIONS
Tetracycline drugs are important in humanmedicine, as well as in veterinary medicine. How-ever, assessing each of several specific suggestedthreats to human health from use of tetracyclinesas STAs—both indirect threats, from co-selectionof MRSA or MDR Salmonella; and direct threats,from infection of human patients with tetracycline-resistant zoonotic bacteria (Table I)—suggests thatreducing tetracycline use in food animals in the UnitedStates should not be expected to cause any improve-ments in human health or to reduce risks of antibiotic-resistant infections. The best point estimate of thehuman health risk of treatment failures caused byanimal use of tetracycline, based on the evidencereviewed in previous sections, is: zero incrementaltetracycline treatment failures per year in the UnitedStates (to several significant digits). However, thepossibility of a risk that is too small to have beenobserved as yet (e.g., on the order of 0 to 1.75 ×10−11 excess treatment failures per 70-year lifetime,due to a tetracycline-resistant infection caused by an-imal use of tetracycline) cannot be ruled out.
In part, the extremely small or zero risk of ex-cess treatment failures is due to the fact that tetra-
cyclines are not used to treat food-borne zoonotic ill-nesses such as campylobacteriosis, salmonellosis, orE. coli infections. In part, it is because, other thanacne, there are few or no resistance-related treatmentfailures of tetracycline prescriptions used in humanmedicine in the United States (see Table I). For acne,human use suffices to maintain a reservoir of resis-tant P. acnes bacteria (as shown by historical ex-perience in Europe), and there is no evidence thattetracycline use in animals contributes to this pool.Finally, although tetracycline resistance is commonin both MRSA and MDR Salmonella, it appears fromrecent data (Gebreyes et al., 2006), that MDR cas-settes that include tetracycline resistance are selectedby other pressures, and not by tetracycline use per se.Time series analyses indicate that both community-acquired and hospital-acquired MRSA are driven byuse of human antibiotics in hospitals.
Potential human health benefits from contin-ued use of tetracyclines and other antibiotics infood animals cannot easily be quantified. Althoughantibiotic-free (ABF) and organic farms have beenreported to have higher prevalence rates than con-ventional farms of bacteria such as Campylobacter inpoultry (Luangtongkum et al., 2006) and Salmonellain pigs (Gebreyes, 2008) (but lower prevalence ofCampylobacter in pigs, Hurd et al., 2008), it is notyet known to what extent, if any, these bacteria affecthuman health. Moreover, resistance rates for severalantibiotics and bacteria are higher in bacteria iso-lated from conventional farms than in bacteria fromABF farms (ibid). For some other (nontetracycline)antibiotics, a comparison of these two potential ef-fects (smaller microbial loads, increased proportionof resistant bacteria) suggests that the potential hu-man health benefits due to reduced bacterial loadsmay outweigh by more than 1,000-fold the potentialrisks due to increased resistance (Cox and Popken,2002, 2004), but similar calculations cannot be madefor tetracycline STAs, as tetracycline is not used totreat zoonotic illnesses in the United States.
In Europe, growth promotion STA bans wereinitially estimated to have increased mortality ratesin pig production, associated with scouring and pro-liferative enteritis, by 10%–15%, and these increasespersisted for years (Hughes and Heritage, 2004; Coxand Ricci, 2008). If morbidity rates in pigs increasedby a similar percentage due to withdrawal of tetracy-clines and other STAs, then the initial results of Hurdet al. (2008), showing a significant positive correla-tion between decreased animal health (indicated bysubclinical lesions) and increased microbial loads of
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 453
Campylobacter and Enterococcus on carcasses, mightprovide a causal mechanism by which reducing an-imal antibiotic use could account for some of theobserved increases in campylobacteriosis (Lawley,2007) and other zoonotic infections in Europe. Tomake this (or other) causal conjecture less specula-tive, it would be desirable to carry out more detailedtime series analyses of changes in illness rates, mi-crobial loads, and resistance fractions in isolates fromanimals, food, and people, as data become available.Denmark has led the way in collecting and publishingsuch data, and similar data are now being collected inother countries.
Without speculating further on causal mecha-nisms and potential human health benefits from con-tinued use of animal antibiotics, it seems clear thatthere is no empirical support for concerns that con-tinued use of tetracycline STAs in the United Stateswill increase human health risks due to tetracycline-resistant illnesses. Some salient threats (e.g., fromterrorist-produced tetracycline-resistant anthrax, orof prolonged courses of oral antibiotic therapy fortetracycline-resistant acne) are unlikely to be af-fected by any changes in animal antibiotic use. (Ter-rorists may use resistant anthrax regardless of ani-mal antibiotic use, while routine use of tetracycline totreat moderate acne cases already provides an amplereservoir of tetracycline-resistant bacteria, for whichother treatments are available.) Other resistancethreats are also minimal because of the absence ofsignificant numbers of tetracycline-resistance-relatedtreatment failures (Table I).
Current urgent political calls to ban tetracyclineand other approved animal antibiotic uses, in an ef-fort to protect human health, are framed in terms ofa mental model of cause and effect in which banswould reduce contamination of family meals with“deadly” antibiotic-resistant pathogens or “super-bugs,” thereby reducing the risk of untreatable food-borne illnesses and deaths (Slaughter, 2008). Our re-view of data for specific illnesses suggests that thismental model has little relation to reality. Food-borne bacteria are not unstoppable or “deadly” ex-cept under rare, serious medical conditions, mostcommonly involving immunocompromised patients,that are not caused by animal antibiotic use. A morerealistic mental model appears to be that, in theUnited States, as in Europe:
1. Reducing animal antibiotic uses that preventanimal illness and promote their health andgrowth would not reduce (but might increase)
microbial loads and human bacterial illnessrates;
2. Reducing STA uses in animals would notreduce antibiotic resistance rates in humaninfections (but might increase them, espe-cially if more antibiotics important in humanmedicine become used for therapy instead ofprevention) (Casewell et al., 2003); and
3. The very real and serious threats of MRSAand MDR “super-bugs” in human patientscan be addressed effectively only by changingthe use of antibiotics in human medicine.
Our analysis suggests that risk management poli-cies based on these principles are far more likely toachieve intended public health benefits than are poli-cies based on the belief or assumption that STA usesin animals contribute significantly to resistance risksin humans (Slaughter, 2008).
ACKNOWLEDMENTS
We thank Alpharma and Phibro Animal Healthfor supporting the research reported here. We thankDr. Jerry Mathers of Alpharma for useful discussionsof recent literature and trends in resistance risks. Theresearch questions addressed, methods and modelsapplied, and conclusions reached are solely those ofthe authors.
REFERENCES
Adams SJ, Cunliffe WJ, Cooke EM. Long-term antibiotictherapy for acne vulgaris: Effects on the bowel floraof patients and their relatives. Journal of InvestigativeDermatology, 1985; 85:35–37. Available at: http://www.nature.com/jid/journal/v85/n1/pdf/5614576a.pdf.
AHFS. AHFS Consumer Medication Information, American So-ciety of Health System Pharmacists. 2008. Available at: http://www.ncbi.nlm.nih.gov/books/bv.fcgi?log$=drug bottom one&rid=medmaster.chapter.a682098.
Akwar HT, Poppe C, Wilson J, Reid-Smith RJ, Dyck M, Wadding-ton J, Shang D, McEwen SA. Associations of antimicrobialuses with antimicrobial resistance of fecal Escherichia colifrom pigs on 47 farrow-to-finish farms in Ontario and BritishColumbia. Can J Vet Res, 2008; 72(2):202–210.
Alban L, Nielsen EO, Dahl J. A human health risk assessmentfor macrolide-resistant Campylobacter associated with the useof macrolides in Danish pig production. Prev Vet Med, 2008;83(2):115–129.
Aldeyab MA, Monnet DL, Lopez-Lozano JM, Hughes CM,Scott MG, Kearney MP, Magee FA, McElnay JC. Modellingthe impact of antibiotic use and infection control practiceson the incidence of hospital-acquired methicillin-resistantStaphylococcus aureus: A time-series analysis. J AntimicrobChemother, 2008; 62(3):593–600.
American Academy of Periodontology. Treatment of plaque-induced gingivitis, chronic periodontitis, and other clinicalconditions. Journal of Periodontology, 2001; 72:1790–1800.
454 Cox and Popken
Angulo FJ, Nargund VN, Chiller TC. Evidence of an associa-tion between use of anti-microbial agents in food animals andanti-microbial resistance among bacteria isolated from hu-mans and the human health consequences of such resistance.J Vet Med B Infect Dis Vet Public Health, 2004; 51(8–9):374–379.
Arsenault J, Letellier A, Quessy S, Morin JP, Boulianne M. Preva-lence and risk factors for Salmonella and Campylobacterspp. carcass contamination in turkeys slaughtered in Quebec,Canada. J Food Prot, 2007; 70(6):1350–1359.
Barza M, Travers K. Excess infections due to antimicrobial re-sistance: The “attributable fraction.” Clin Infect Dis, 2002;34(Suppl 3):S126–S130.
Beck J, Offenbacher S. Potential public health implicationsperiodontal disease and cardiovascular disease. In Proceedingof the Public Health Implications of Chronic Periodontal In-fections in Adults Conference, Atlanta, Georgia, 2003. Avail-able at: http://www.cdc.gov/ORALHEALTH/publications/library/conferences/periodontal infections08.htm, Accessedon April 8–9, 2003.
Beilby J, Marley J, Walker D, Chamberlain N, Burke M; FIESTAStudy Group. Effect of changes in antibiotic prescribing onpatient outcomes in a community setting: A natural experi-ment in Australia. Clin Infect Dis, 2002; 34(1):55–64.
Bell DM, Kozarsky PE, Stephens DS. Conference summary: Clin-ical issues in the prophylaxis, diagnosis, and treatment ofAnthrax. Emerging Infectious Diseases, 2002; 8(2):222–225.Available at: http://www.cdc.gov/ncidod/EID/vol8no2/pdf/01-0521.pdf.
Berge AC, Adaska JM, Sischo WM. Use of antibiotic suscepti-bility patterns and pulsedfield gel electrophoresis to comparehistoric and contemporary isolates of multi-drug-resistantSalmonella enterica subsp. enterica serovar Newport. ApplEnviron Microbial, 2004; 70(1):318–323.
Bosso JA, Mauldin PD. Using interrupted time series analysisto assess associations of fluoroquinolone formulary changeswith susceptibility of gram-negative pathogens and isolationrates of methicillin-resistant Staphylococcus aureus. Antimi-crob Agents Chemother, 2006; 50(6):2106–2112.
Bryan A, Shapir N, Sadowski MJ. Frequency and distribution oftetracycline resistance genes in genetically diverse, nonse-lected and nonclinical Escherichia coli strains isolated fromdiverse human and animal sources. Applied and Environ-mental Microbiology, 2004; 70(4):2503–2507. Available at:http://www.pubmedcentral.nih.gov/picrender.fcgi?tool=pmcentrez&artid=383146&blobtype=pdf.
Casewell M, Friis C, Marco E, McMullin P, Phillips I. The Eu-ropean ban on growth-promoting antibiotics and emergingconsequences for human and animal health. J AntimicrobChemother, 2003; 52(2):159–161.
CDC. Strategies for clinical management of MRSA in thecommunity: Summary of an experts meeting convened bythe Centers for Disease Control and Prevention—March2006. 2006. Available at: http://www.cdc.gov/ncidod/dhqp/pdf/ar/CAMRSA ExpMtgStrategies.pdf.
CDC. Preliminary FoodNet data on the incidence of infec-tion with pathogens transmitted commonly throughfood—10 states, 2007. CDC MMWR Weekly, 2008;57(14):366–370. Available at: http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5714a2.htm.
Chan PA, Wakeman SE, Angelone A, Mermel LA. In-vestigation of multi-drug resistant microbes in retailmeats. Food, Agriculture, & Environment, 2008; 6(3–4):71–75. Available at: http://www.world-food.net/scientficjournal/2008/issue3/abstracts/abstract15.php.
Chander Y, Gupta S, Goyal S, Kumar K, Murray H. Preva-lence of antibiotic resistant bacteria on swine farms. Proc.ASA/CSSA/SSSA International Annual Meetings. November12–16, 2006.
Chee-Sanford JC, Aminov RI, Krapac IJ, Garrigues-JeanjeanN, Mackie RI. Occurrence and diversity of tetracyclineresistance genes in lagoons and groundwater underlyingtwo swine production facilities. Applied EnvironmentalBiology, 2001; 67(4):1494–1502. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=92760.
Christensen B, Sommer H, Rosenquist H, Nielsen N. RiskAssessment on Campylobacter jejuni in Chicken Prod-ucts, 1st ed. 2001. Available at: http://www.foedevaredirektoratet.dk/NR/rdonlyres/emmun3ukyfeweov2kt5f72bngkszsy3g4tfwzd2do4qlyn7my2bseotxcrdyknz7bndtoe2p6pkzmszbqabatvo4cje/rapport jan2001.pdf.
Cox LA Jr. Does concern-driven risk management provide a vi-able alternative to QRA? Risk Analysis, 2007; 27(1):27–43.
Cox LA Jr. Some limitations of a proposed linear model for an-timicrobial risk management. Risk Analysis, 2005; 25(6):1327–1332.
Cox LA Jr, Babayev D, Huber W. Some limitations of qualitativerisk rating systems. Risk Analysis, 2005; 25(3):651–662.
Cox LA Jr, Popken DA. Quantifying potential human health im-pacts of animal antibiotic use: enrofloxacin and macrolides inchickens. Risk Anal, 2006; 26(1):135–146.
Cox LA, Popken DA. Quantifying human health risks from vir-giniamycin used in chickens. Risk Analysis, 2004; 24(1):271–288.
Cox LA, DA Popken. A simulation model of human health risksfrom chicken-borne Campylobacter jejuni. Technology, 2002;9:55–84.
Cox LA Jr, Popken DA, Mathers J. Human health risk assessmentof penicillin/aminopenicillin resistance in enterococci due topenicillin use in food animals. Risk Analysis, 2009; 29(6):796–805.
Cox LA Jr, Ricci PF. Causal regulations vs. political will: Whyhuman zoonotic infections increase despite precautionarybans on animal antibiotics. Environment International, 2008May;34(4):459–475.
Del Rosso JQ, Kim G. Optimizing use of oral antibiotics in acnevulgaris. Dermatol Clin, 2009; 27(1):33–42.
Dunne EF, Fey PD, Kludt P, et al. Emergence of domestically ac-quired ceftriaxone-resistant Salmonella infections associatedwith AmpC bet-lactamase. JAMA, 2000; 284(24):3151–3156.Available at: http://jama.ama-assn.org/cgi/content/full/284/24/3151.
Eady AE, Cove JH, Layton AM. Is antibiotic resistance in cuta-neous Propionibacteria clinically relevant? Am J Clin Derma-tol, 2003; 4(12):813–831.
Engemann CA, Keen PL, Knapp CW, Hall KJ, Graham DW. Fateof tetracycline resistance genes in aquatic systems: Migrationfrom the water column to peripheral biofilms. Environ Sci,Technol, 2008; 42(14):5131–5136.
EPI-NEWS. DANMAP 2007: Antimicrobial consumption & resis-tance. EPI-NEWS—National Surveillance of CommunicableDiseases 47 [serial online], 2008. Available at: http:// www.ssi.dk/sw60246.asp.
FAO/WHO/OIE. Joint FAO/WHO/OIE Expert Meeting onCritically Important Antimicrobials. Report of a meetingheld in FAO, Rome, Italy, November 26–30 2007. FAO,Rome, Italy, and WHO, Geneva, Switzerland, 2008. Avail-able at: http://www.fao.org/ag/agn/agns/files/Prepub ReportCIA.pdf.
FDA-CVM. (US Food and Drug Administration, Center for Vet-erinary Medicine). Draft risk assessment of streptograminresistance in Enterococcus faecium attributable to the useof streptogramins in animals, 2004, November 23, 2004.Available at: http://www.fda.gov/cvm/antimicrobial/SREFRA FinalDraft.pdf.
FDA-CVM. (US Food and Drug Administration, Center for Vet-erinary Medicine). Guidance for Industry # 152: Evaluatingthe safety of antimicrobial new animal drugs with regard to
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 455
their microbiological effects on bacteria of human health con-cern. US Dept of Health and Human Services, Food and DrugAdministration, Center for Veterinary Medicine, 2003. Avail-able at: http://www.fda.gov/cvm/Guidance/fguide152.pdf.
FDA-CVM. Risk assessment on the human health impact offluoroquinolone resistant campylobacter associated with theconsumption of chicken. Food and Drug Administration—Center for Veterinary Medicine, 2001. Available athttp://www.fda.gov/AnimalVeterinary/SafetyHealth/RecallsWithdrawals/ucm042019.htm. Last accessed on 12/21/2009.
Feller W. An Introduction to Probability Theory and its Applica-tions, Volume I. 3rd ed. New York: Wiley, 1968.
Fritz JM, Woeltje KF. Mycobacterium Fortuitum, 2007. Availableat: http:// www.emedicine.com/med/topic1534.htm.
Gebreyes WA. Salmonella in Swine. FSRRN Journal, 2008; IV(1).Available at: http://www.fsrrn.net/modules/content/index.php?id=157.
Gebreyes WA, Thakur S, Morrow WE. Comparison of preva-lence, antimicrobial resistance, and occurrence of multidrug-resistant Salmonella in antimicrobial-free and conventionalpig production. J Food Prot, 2006; 69(4):743–748.
Gebreyes WA, Davies PR, Turkson P-K, Morrow WEM, FunkJA, Altier C, Thakur S. Characterization of antimicrobial-resistant phenotypes and genotypes among Salmonella entericrecovered from pigs on farms, from transport trucks, and frompigs after slaughter. International Journal of Food Protection,2004; 67(4):698–705.
Goossens H, Ferech M, Coenen S, Stephens P; The EuropeanSurveillance of Antimicrobial Consumption Project Group.Comparison of outpatient systemic antibacterial use in 2004 inthe United States and 27 European countries. Clinical Infec-tious Diseases, 2007; 44:1091–1095. Available at: http://www.journals.uchicago.edu/doi/pdf/10.1086/512810.
Gould IM. A review of the role of antibiotic policies in the con-trol of antibiotic resistance. J Antimicrob Chemother, 1999Apr;43(4):459–465.
Graham D. Mechanisms of tetracycline resistance development inthe environment as detected by real-time PCR. Presentationat the US EPA Meeting on Pharmaceuticals in the Environ-ment, Las Vegas, NV, August 23–25, 2005. Abstract availableat: http://es.epa.gov/ncer/publications/workshop/8-23-2005/abstract/graham.html.
Gray JT, Hungerford LL, Fedorka-Cray PJ, Headrick ML.Extended-spectrum-cephalosporin resistance in Salmonellaenteric isolates of animal origin. Antimicrob AgentsChemother, 2004; 48(8):3179–3181. Available at: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=478523.
Greenstein G. Nonsurgical periodontal therapy in 2000: A lit-erature review. J Am Dent Assoc, 2000; 131(11):1580–1592.Available at: http://jada.ada.org/cgi/content/full/131/11/1580.
Guay DR. Topical clindamycin in the management of acne vul-garis. Expert Opin Pharmacother, 2007; 8(15):2625–264.
Hayes Jr, Jensen H. Lessons from the Danish ban on feed-gradeantibiotics. Briefing paper O3-BP 41. Center for Agriculturaland Rural Development, Iowa State University, 2003. Avail-able at: http://www.card.iastate.edu/publications/DBS/PDFFiles/03bp41.pdf.
Heike S, Stoob K, Hamscher G, Smit E, Seinen W. Tetracyclinesand tetracycline resistance in agricultural soils: Microcosmand field studies. Microbial Ecology, 2006; 51(3):267–276.
Helms RA, Herfindal ET, Quan DJ, Gourley DR. Textbookof Therapeutics: Drug and Disease Management. LippincottWilliam & Wilkins, 2006.
Herron-Olson L, Fitzgerald JR, Musser JM, Kapur V. Molecu-lar correlates of host specialization in Staphylococcus aureus.PLoS ONE, 2007; 2(10):e1120.
Hughes P, Heritage J. Antibiotic growth promoters in foodanimals, 2004. Available at: http://www.fao.org/docrep/007/y5159e/y5159e08.htm.
Hurd HS, Brudvig J, Dickson J, Mirceta J, Polovinski M, MatthewsN, Griffith R. Swine health impact on carcass contamina-tion and human foodborne risk. Public Health Rep, 2008;123(3):343–351.
Hurd HS, Malladi S. A stochastic assessment of the public healthrisks of the use of macrolide antibiotics in food animals. RiskAnalysis, 2008; 28(3):695–710.
Inglis GD, McAllister TA, Busz HW, Yanke LJ, Morck DW,Olson ME, Read RR. Effects of subtherapeutic adminis-tration of antimicrobial agents to beef cattle on the preva-lence of antimicrobial resistance in Campylobacter jejuniand Campylobacter hyointestinalis. Appl Environ Microbiol,2005; 71(7):3872–3881.
Kang M-S, Besser TE, Hancock DD, Porwollik S, McCelland M,Call DR. Identification of specific gene sequences conservedin contemporary epidemic strains of Salmonella enteric. Ap-plied and Environmental Microbiology, 2006; 72:6938–6947.Available at: http://aem.asm.org/cgi/reprint/72/11/6938.
Karatzas KGG, Randall LP, Webber M, Piddock LJV, HumphreyTJ, Woodward M, Coldham NG. Phenotypic and proteomiccharacterization of multiply antibiotic-resistant variants ofSalmonella enterica serovar Typhimurium selected followingexposure to disinfectants. Applied and Environmental Micro-biology, 2008; 74(5):1508–1516.
KAW (Keep Antibiotics Working) website. http://www.keepantibioticsworking.org/new/top act.cfm (This is KAW’s web-site, and summarizes the top actions that they advocate.) Seealso http://www.keepantibioticsworking.com/new/resourceslibrary.cfm?refID=36396), 2009. Last retrieved 11/20/2009.
KAW (Keep Antibiotics Working). MRSA and animal agricul-ture: A need to understand the public health impact, 2008.Available at: http://www.keepantibioticsworking.com/new/indepth mrsa.cfm.
Kertzman MGM, Smeets JGE, Boukes FS, Goudswaard AN.Summary of the practice guideline “Acne” (second revision)from the Dutch College of General Practitioners] Ned Tijd-schr Geneeskd, 2008; 152(22):1271–1275. Available at: http://lib.bioinfo.pl/meid:57443.
Klein NC, Cunha BA. Tetracyclines. Med Clin North Am, 1995;79(4):789–801.
Kleinfelder JW, Muller RF, Lange DE. Antibiotic susceptibilityof putative periodontal pathogens in advanced periodontitispatients. J Clin Periodontol, 1999; 26(6):347–351.
Klevens RM et al. Invasive methicillin-resistant Staphylo-coccus aureus infections in the United States. JAMA,2007; 298(15):1763–1771. Available at: http://www.cdc.gov/ncidod/dhqp/ pdf/ar/InvasiveMRSA JAMA2007.pdf.
Langner A, Chu A, Goulden V, Ambroziak M. A randomized,single-blind comparison of topical clindamycin + benzoyl per-oxide and adapalene in the treatment of mild to moder-ate facial acne vulgaris. Br J Dermatol, 2008; 158(1):122–129.
Lawley R. Campylobacter takes over as number one. Food SafetyWatch 2007, 2007. Available at: http://www.foodsafetywatch.com/public/471.cfm?sd=2.
Lewis HC, Molbak K, Reese C, Aarestrup FM, Selchau M, SorumM, Skov R. Pigs as a source of methicillin-resistant Staphylo-coccus aureus infections in humans, Denmark. Emerg InfectDis, 2008; 14(9):1383–1389. Available at: http://www.cdc.gov/eid/content/14/9/pdfs/1383.pdf.
Lorian V. Antibiotics in Laboratory Medicine. LippincottWilliams & Wilkins, 2005.
Luangtongkum T, Morishita TY, Ison AJ, Huang S, McDermottPF, Zhang Q. Effect of conventional and organic productionpractices on the prevalence and antimicrobial resistance ofCampylobacter spp. in poultry. Appl Environ Microbiol, 2006;72(5):3600–3607.
Lundergan WP. Periodontal antibiotic/antimicrobial therapy. InHall Water B (ed). Critical Decisions in Periodontology.
456 Cox and Popken
London: BC Decker Inc, 2003. Available at: http://dentistry.tums.ac.ir/Files/lib/My%20Web%20Sites/CD PER%20(E)/docs/ch051.pdf.
MacDonald JM, McBride WD. The transformation of U.S. live-stock agriculture: Scale, efficiency, and risks. Economic Re-search Service/USDA Economic Information Bulletin No.(EIB-43) January 2009, 2009. Available at: http://www.ers.usda.gov/Publications/EIB43.
MacKenzie FM, Lopez-Lozano JM, Monnet DL, Stuart D,Beyaert A, Wilson R, Gould IM. Temporal relationship be-tween prevalence of methicillin-resistant Staphylococcus au-reus (MRSA) in one hospital and prevalence of MRSA in thesurrounding community: A time-series analysis. J Hosp Infect,2007; 67(3):225–231.
McKeage K, Keating GM. Clindamycin/benzoyl peroxide gel(BenzaClin): A review of its use in the management of acne.Am J Clin Dermatol, 2008; 9(3):193–204.
Merck. Acne Vulgaris. The Merck Manuals Online Medical Li-brary, 2008. Available at: http://www.merck.com/mmpe/sec10/ch111/ch111b.html#sec10-ch111-ch111b-182.
Mirbagheri SA, Hasibi M, Abousari M, Rashidi A. Triple,standard quadruple and ampicillin-sulbactam-based quadru-ple therapies for H pylori eradication: A comparativethree-armed randomized clinical trial. World Journal ofGastroenterology, 2006; 12(30):4888–4891. Available at:http://www.wjgnet.com/1007-9327/12/4888.asp.
Mølbak K. Spread of resistant bacteria and resistance genes fromanimals to humans—The public health consequences. J VetMed B Infect Dis Vet Public Health, 2004; 51(8–9):364–369.
Monnet DL, Frimodt-Moller N. Antimicrobial-drug use andmethicillin-resistant Staphylococcus aureus. Letter in Emerg-ing Infectious Diseases, 2001; 7(1) [serial online]. Available at:http://www.cdc.gov/ncidod/eid/vol7no1/monnet letter.htm.
Monnet DL, MacKenzie FM, Lopez-Lozano JM, Beyaert A, Ca-macho M, Wilson R, Stuart D, Gould IM. Antimicrobialdrug use and methicillin-resistant Staphylococcus aureus, Ab-erdeen, 1996–2000. Emerg Infect Dis, 2004; 10(8):1432–1441.
Morgan M. Methicillin-resistant Staphylococcus aureus and an-imals: Zoonosis or humanosis? Journal of AntimicrobialChemotherapy, 2008; 62(6):1181–1187.
NARMS (National Antimicrobial Resistance Monitoring System).NARMS veterinary isolates, 2009. Available at: http://www.ars.usda.gov/Main/docs.htm?docid=6750&page=4#Tables.
NIH. E. coli enteritis. US National Library of Medicine (Med-linePlus), National Institute of Health, 2008. Available at:http://www.nlm.nih.gov/medlineplus/ency/article/000296.htm#Treatment.
Olsvik B, Tenover FC. Tetracycline resistance in periodontalpathogens. Clinical Infectious Diseases, 1993; 16(Sup 4):S310–S313.
Ozolins M, Eady EA, Avery A, Cunliffe WJ, O’Neill C, SimpsonNB, Williams HC. Randomised controlled multiple treatmentcomparison to provide a cost-effectiveness rationale for theselection of antimicrobial therapy in acne. Health Technol As-sess, 2005; 9(1):iii–212.
Patchanee P, Zewde BM, Tadesse DA, Hoet A, Gebreyes WA.Characterization of multidrug-resistant Salmonella entericaserovar Heidelberg isolated from humans and animals. Food-borne Pathog Dis, 2008; 5(6):839–851.
Peak N, Knapp CW, Yang RK, Smith MM Marilyn S, AgaDS, Graham DW. Abundance of six tetracycline resis-tance genes in wastewater lagoons at cattle feedlots withdifferent antibiotic use strategies. Environmental Microbi-ology, 2007; 9(1):143–151. Available at: http://www.ingentaconnect.com/content/bsc/emi/2007/00000009/00000001/art00022;jsessionid=2f0n.
Pew. Pew applauds introduction of preservation of antibi-otics for Medical Treatment Act, 2009. http://www.save
antibiotics.org/newsroom/pr 17march2009.html. Last re-trieved 11/20/2009.
Phillips I. Withdrawal of growth-promoting antibiotics in Europeand its effects in relation to human health. Int J AntimicrobAgents, 2007; 30(2):101–107.
Pu S, Han F, Ge B. Isolation and characterization ofmethicillin-resistant Staphylococcus aureus from Louisianaretail meats. Applied and Environmental Microbiol-ogy, 2009; 75(1):265–267. Available at: http://www.ncbi.nlm.nih.gov/pubmed/18978079.
Pugh DM. The EU precautionary bans of animal feed additive an-tibiotics. Toxicol Lett, 2002; 128(1–3):35–44.
Ricci V, Piddock LJV. Ciprofloxacin selects for multidrug re-sistance in Salmonella enteric serovar Typhimurium medi-ated by at least two different pathways. Journal of An-timicrobial Chemotherapy, 2009; 63(5):909–916. Available at:http://jac.oxfordjournals.org/cgi/content/abstract/63/5/909.
Roberts M. Tetracycline therapy: Update. Clinical InfectiousDiseases, 2003; 36:462–467. Available at: http://www.journals.uchicago.edu/doi/pdf/10.1086/367622.
Rodrigues RM, Goncalves C, Souto R, Feres-Filho EJ, Uzeda M,Colombo AP. Antibiotic resistance profile of the subgingivalmicrobiota following systemic or local tetracycline therapy. JClin Periodontol, 2004; 31(6):420–427.
Romanowicz M, Stephenson JJ, Del Rosso JQ, Lenhart G.Healthcare utilization and costs of patients with rosaceain an insured population. Journal of Drugs in Derma-tology, 2008; 7(1):41–49. Available at: http://findarticles.com/p/articles/mi m0PDG/is /ai n24258971.
Ross JI, Snelling AM, Carnegie E, Coates P, Cunliffe WJ, Bet-toli V, Tosti G, Katsambas A, Galvan Perez Del Pulgar JI,Rollman O, Torok L, Eady EA, Cove JH. Antibioticresistantacne: Lessons from Europe. Br J Dermatol, 2003; 148(3):467–478.
Russell SSF. General researchers follow strain of drug-resistant bacteria. San Francisco Chronicle, 15 Jan 2008,2008. Available at: http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2008/01/15/MNUKUDB6D.DTL&feed=rss.news.
Sack DA, Kaminsky DC, Sack RB, Itotia JN, Arthur RR,Kapikian AZ, Orskov F, Orskov I. Prophylactic doxycyclinefor travelers’ diarrhea. Results of a prospective double-blindstudy of Peace Corps volunteers in Kenya. N Engl J Med,1978; 298(14):758–763.
Schneider A. Potentially fatal bacteria found in pigs, farmwork-ers. Seattle pi.com 8 Jun 2008, 2008. Available at: http://seattlepi.nwsource.com/local/366301 pigmrsa09.html.
Simmons S. In the news 2006–2007: Bipartisan Bill curtailsroutine use of antibiotics in animal feed. Vermont Jour-nal of Environmental Law, 2007. Available at: http://www.vjel.org/news/NEWS100074.html. Last retrieved 11/20/2009.
Simonart T, Dramaix M, De Maertelaer V. Efficacy of tetracy-clines in the treatment of acne vulgaris: A review. Br J Der-matol, 2008; 158(2):208–216.
Singer RS, Cox LA Jr., Dickson JS, Hurd HS, Phillips I,Miller, GY. Modeling the relationship between food animalhealth and human foodborne illness. Prev Vet Med, 2007;79(2–4):186–203.
Skidmore R, Kovach R, Walker C, et al. Effects ofsubantimicrobial-dose doxycycline in the treatment ofmoderate acne. Arch Dermatol, 2003; 139:459–464. Avail-able at: http://archderm.ama-assn.org/cgi/reprint/139/4/459?ijkey=c77757416209b8f0b914f7f7a95bb2725d4fdf0b.
Slaughter LM. Chairwoman Slaughters statement on AgriculturalSubcommittee hearing on animal health, antibiotics. Sep 25th,2008, 2008. Available at: http://www.ncrlc.com/antibiotics-agri page2.html.
Smith JL, Drum DJV, Daj Y, Kim JM, Sanchez S, Maurer JJ,Hofacre CL, Lee MD. Impact of antimicrobial usage on an-timicrobial resistance in commensal Eshcerichia coli strains
Assessing Potential Human Health Hazards and Benefits from Subtherapeutic Antibiotics 457
colonizing broiler chickens. Appl. and Environ. Microbiol,2007; 73:1404–1414.
Smith TC. Dr. Tara Smith explores zoonotic MRSA. Cen-ter for Emerging Infectious Diseases Newsletter Vol 9,Summer 2008, 2008b. Available at: http://www.public-health.uiowa.edu/CEID/Center news/CEID news Sept 18 2008.pdf.
Smith TC, Male MJ, Harper AL, Kroeger JS, Tinkler GP, MoritzED, Capuano AW, Herwaldt LA, Diekema DJ. Methicillin-resistant Staphylococcus aureus (MRSA) strain ST398 ispresent in midwestern U.S. swine and swine workers. PLoSONE, 2009; 4(1):e4258.
Smith TC, Male MF, Harper AL, Moritx-Kolev E, Kroeger JS,Dickema DJ, Herwaldt LA. Isolation of methicillin-resistantStaphylococcus aereus (MRSA) from swine in the midwesternUnited States. Presentation at the International Conferenceon Emerging Infectious Disease (ICEID), Atlanta, GA,Mar 16–19, 2008. Available at: http://www.public-health.uiowa.edu/CEID/Posters/Smith%20ICEID%20presentation%203-18-08.pdf.
Stern NJ, Robach MC. Enumeration of Campylobacter spp. inbroiler feces and in corresponding processed carcasses. J FoodProt, 2003; 66(9):1557–1563.
Stern RS. Medication and medical service utilization for acne1995–1998. J am Acad Dermatol, 2000; 43(6):1042–1048.
Swanson JK. Antibiotic resistance of Propionibacteriumacnes in acne vulgaris. Dermatology Nursing, 2003;15(40):359–362. Available at: http://findarticles.com/p/articles/mi hb6366/is 4 15/ai n29025578/print?tag=artBody;col1.
Swenson JM, Wallace RJ, Silcox VA, Thornsberry C. Antimi-crobial susceptibility of five subgroups of Mycobacteriumfortuitum and Mycobacterium chelonae. AntimicrobialAgents and Chemotherapy, 1985; 28(6):807–811. Available at:http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=180333.
Tacconelli E, De Angelis G, Cataldo MA, Pozzi E, Cauda R.Does antibiotic exposure increase the risk of methicillin-resistant Staphylococcus aureus (MRSA) isolation? A system-atic review and meta-analysis. J Antimicrob Chemother, 2008;61(1):26–38.
Tan AW, Tan HH. Acne vulgaris: a review of antibiotic therapy.Expert Opin Pharmacother, 2005; 6(3):409–418.
Thakur S, Tadesse DA, Morrow M, Gebreyes WA. Occurrenceof multidrug resistant Salmonella in antimicrobial-free (ABF)swine production systems. Vet Microbiol, 2007; 125(3–4):362–367.
Thevarajah S, Balkrishnan R, Camacho FT, Feldman SR, Fleis-cher AB Jr. Trends in prescription of acne medication in theUS: Shift from antibiotic to non-antibiotic treatment. J Der-matolog Treat, 2005; 16(4):224–228.
UCS (Union of Concerned Scientists). Tell Congress: Keepantibiotics working, 2009. https://secure3.convio.net/ucs/site/Advocacy?cmd=display&page=UserAction&id=1924 (Unionof Concerned Scientists call for letters to members ofCongress). Last retrieved 11/20/2009.
Valtonen MV, Valtonen VV, Salo OP, Makela PH. The ef-fect of tetracycline treatment for acne vulgaris on the oc-
currence of R factors in the intestinal flora of man. BritishJournal of Dermatology, 1976; 4:71–74. (published online2006).
Van Immerseel F, Rood JI, Moore RJ, Titball RW. Rethinkingour understanding of the pathogenesis of necrotic enteritis inchickens. Trends Microbiol, 2009; 17(1):32–36.
Van Loo IHM, Diederen BMW, Savelkout PHM, Wouden-berg JHC, Roosendaal R, van Belkum A, Lemmens-den Toom N, Verhulst C, van Keulen HJ, KluytmansJAJW. Methicillin-resistant Staphylococcus aureus inmeat products, the Netherlands. Emerging Infectious Dis-eases, 2007; 13(11) [serial on the Internet]. Available at:http://www.cdc.gov/EID/content/13/11/1753.htm.
Varma JK, Marcus R, Stenzel SA, Hanna SS, Gettner S, An-derson BJ, Hayes T, Shiferaw B, Crume TL, Joyce K,Fullerton KE, Voetsch AC, Angulo FJ. Highly resistantSalmonella Newport-MDRAmpC transmitted through thedomestic US food supply: A FoodNet case-control study ofsporadic Salmonella Newport infections, 2002–2003. J InfectDis, 2006; 194(2):222–230.
Vierikko A, Hanninen M-J, Siitonen A, Ruutu P, Rautelin H.Domestically acquired Campylobacter infections in Finland.Emerg Infect Dis, 2004; 10(1) [serial online]. Available at:http://www.cdc.gov/ncidod/EID/vol10no1/03-0636.htm.
Villedieu A, Diaz-Torres ML, Hunt N, McNab R, SprattDA, Wilson M, Mullany P. Prevalence of tetracyclineresistance genes in oral bacteria. Antimicrobial Agentsand Chemotherapy, 2003; 47(3):878–882. Available at:http://aac.asm.org/cgi/content/full/47/3/878.
Weber SG, Gold HS, Hooper DC, Karchmer AW, Carmeli Y. Flu-oroquinolones and the risk for methicillin-resistant Staphylo-coccus aureus in hospitalized patients. Emerg Infect Dis, 2003;9(11):1415–1422.
Weigel RM, Isaacson RE. Patterns of transmission of antibioticresistant Salmonella to swine. Reseach Report 99-046, Na-tional Pork Producers Council, 2004. Available at: http://www.pork.org/PorkScience/Documents/REPORT%2004-99-046-Weigel-U%20of%20IL.pdf.
World Health Organization. Drug-resistant Salmonella. FactSheet # 139, revised April 2005, 2005. Available at: http://www.who.int/mediacentre/factsheets/fs139/en/.
World Health Organization. Impacts of antimicrobial growthpromoter termination in Denmark, 2003. Available at:http://whqlibdoc.who.int/hq/2003/WHO CDS CPE ZFK 2003.1.pdf.
Wulf MW, Sørum M, van Nes A, Skov R, Melchers WJ, KlaassenCH, Voss A. Prevalence of methicillin-resistant Staphylococ-cus aureus among veterinarians: An international study. ClinMicrobiol Infect, 2008; 14(1):29–34.
Zaidi MB, Leon V, Canche C, Perez C, Zhao S, Hubert SK,Abbott J, Blickenstaff K, McDermott PF. Rapid andwidespread dissemination of multidrug-resistant blaCMY-2 Salmonella Typhimurium in Mexico. J AntimicrobChemother, 2007; 60(2):398–401.
Zeichner SL. Tetracycline update. Pediatrics in Review, 1998;19:32.
