July 11, 2003
Kent D. McClure, D.V.M., J.D.
General Counsel
Animal Health Institute
1325 G Street, N.W., Suite 700
Washington, D.C. 20005-3104
RE: Appeal of Request for Correction of Information; letter and accompanying materials dated March 30, 2003, to Julie Gerberding, MD, MPH, Director of the Centers for Disease Control and Prevention (CDC) and Administrator for the Agency for Toxic Substances and Disease Registry (ATSDR) from Kent D. McClure, DVM, JD, General Counsel for the Animal Health Institute.
Dear Dr. McClure:
I am responding, on behalf of the Centers for Disease Control and Prevention (CDC), to your March 30, 2003 Appeal of the Denial by CDC of your Request for Correction of Information disseminated by CDC. In making this response, CDC has considered your December 6, 2002 Request, as amended on January 8, 2003, CDC's Denial of that request, your March 30, 2003 Appeal and the information you provided at our May 13, 2003 meeting. Responses to your requests are provided in the order that you presented them.
REQUEST FOR CORRECTION 1: You requested the correction or deletion of CDC's statements that a person with Campylobacterinfection that is resistant to fluoroquinolone will have an extended illness or two days of diarrhea more than a person that is infected withCampylobacter that is susceptible to fluoroquinolones. In support of this request, Animal Health Institute (AHI) contends that CDC directly (e.g. [1])* and by quoting a study by Smith et al, published in the New England Journal of Medicine (2) attribute excess days of illness to the fluoroquinolone resistance of the infecting strains, that AHI contends actually occur because of foreign travel. AHI contends that foreign travel is statistically significant and should have been considered by CDC as an important confounder. AHI cites a peer-reviewed journal article by L. A. Cox (3) in support of this request.
CDC Response to the appeal: CDC must deny this request.
As noted before, the CDC/FoodNet case-control study of campylobacteriosis shows that persons infected with Campylobacter that is resistant to fluoroquinolones have a longer illness than do persons infected with susceptible strains of Campylobacter (1). The report published by Cox, et al, in 2002, on which you rely, notes in one sentence, unaccompanied by data, that he found no significant association between fluoroquinolone resistance in domestic adult cases and duration of diarrhea ([3], page 3S33). As described below, CDC does not agree with the analytic approach that he used.
CDC agrees that it is important to control for major confounders in the statistical analysis. In doing so, it is important to review a series of potential confounders, starting with those that are most obvious and biologically plausible. Because the treatment of diarrheal illness with antidiarrheal medications may well be associated with the reported duration of diarrhea, CDC scientists thought it was most important to control for such treatment effects. In analyses that do control for such treatment effects, there is a consistent and robust association between fluoroquinolone resistance and greater duration of diarrhea across a variety of analytic assumptions and models. When these analyses also include foreign travel as a variable, the inclusion of foreign travel does not change the consistent association observed between fluoroquinolone resistance and duration of diarrhea; foreign travel is not consistently or strongly associated with longer duration of illness, nor does it confound the observation that resistance is associated with longer duration illness. Cox's analysis of the same data set has not controlled for the treatment effect. CDC believes it is essential to do so. Failure to include the effect of treatment would leave a major associated factor uncontrolled. Thus, as the findings remain similar in analyses which also include travel (and thus control for that effect), the conclusions reached are justified.
In response to AHI's supposition that there are important geographic differences among Campylobacter jejuni strains that mandate the consideration of foreign travel as a confounder, CDC remains unaware of demonstrated differences in the virulence of Campylobacter jejunistrains among countries. The publication by Professor Coker, cited in your appeal in support of AHI's supposition, summarizes major epidemiological and clinical differences that have been described comparing Campylobacter infections in developing countries to those in developed countries, but does not attribute these to differences in the bacterial strains themselves (4). Rather Professor Coker notes that infections in inhabitants of the developing world often appear to be shorter and less severe than in the developed world, likely because these individuals in the developing world have pre-existing immunity. This prior immune experience with Campylobacter is a more likely explanation for the observed clinical differences, than inherent differences in the virulence of bacterial strains in different parts of the world. Travelers from developed countries, without the immunity generated by early and frequent exposure, get an illness similar to what is experienced among those who acquire infection without leaving their developed country.
Some cases of fluoroquinolone resistant Campylobacter infection occur in persons traveling to other parts of the developed world, not to the developing world (2). Because no clear virulence differences separate strains by country or site of acquisition, it is not necessary to exclude foreign-travel associated cases from an assessment of the clinical impact of microbial characteristics such as resistance on duration of illness.
REQUEST FOR CORRECTION 2: AHI requests that CDC correct its statement that the data show that the percentage of fluoroquinolone resistant Campylobacter isolated from people in the United States is dramatically rising and that such rise is attributable to the use of fluoroquinolone agents in poultry. AHI supports its request by stating that there were high levels of resistance in the 1970s and 80s before the introduction of fluoroquinolone agents in poultry production; that the National Antimicrobial Resistance Monitoring System (NARMS) database relied upon by CDC for these statements is not representative of the U.S. population and is flawed and that CDC does not justify the causal attribution of any temporal trends in resistance risk to use of fluoroquinolone agents in poultry as opposed to other sources, such as contaminated water.
CDC Response to the appeal: For the reasons stated below, CDC must deny this request.
CDC findings do not support your assertion that there were high levels of fluoroquinolone resistance before the introduction of enrofloxacin (Baytril), a fluoroquinolone approved for use in poultry in 1995. Fluoroquinolone resistance was not identified in the predominant species ofCampylobacter, C. jejuni, in the CDC Sentinel County Study conducted in 1989-1990 ([5] and CDC unpublished data). Nachamkin similarly observed no fluoroquinolone resistance in C. jejuni strains isolated in Philadelphia from 1982 through 1992 (6). Using an experimental method for laboratory isolation, Kiehlbauch reported 12% of C. jejuni/coli strains isolated in Wisconsin from 1992-1995 were not susceptible to fluoroquinolone agents, without mentioning a trend; travel status for these patients was also not reported (7). Smith in Minnesota reported that the first ciprofloxacin resistant strain there was identified in 1993, one year after their survey began, though travel status is not reported for this patient (2). The actual exposures to contaminated food or water that resulted in travel-associated infections is uncertain, but it is noteworthy that fluoroquinolone agents began to be used for agricultural purposes in other countries years before this occurred in the United States. For example, fluoroquinolone resistance in Campylobacter isolated from humans appeared concurrent with the use of enrofloxacin in poultry in the Netherlands in 1987-8 (8).
Fluoroquinolone agents were first licensed for use in poultry in the United States in 1995. Following licensure, the proportion ofCampylobacter strains that were resistant to fluoroquinolones was observed to be higher than observed in the 1989 - 90 CDC Sentinel County Study. Nachamkin reported this from Philadelphia, noting it had increased from 20% in 1995 to 40% by 2001, though travel status of the individuals from whom these isolates came was not reported (9). Smith et al. reported that among those infections tin Minnesota that were not associated with foreign travel, the proportion that were resistant increased from 0.8% in 1996 to 3% in 1998, a significant trend, as evaluated by the chi-square for linear trend, p=0.002) (2). NARMS data shows a trend of increasing proportion of Campylobacter that were resistant between 1997 when this surveillance first began, (and when 12% of Campylobacter jejuni strains were resistant) and 2001 (when 18% ofCampylobacter jejuni strains were resistant), an increase that is statistically significant (10). While some of these cases were acquired during foreign travel, and are thus associated with exposures in other countries, two studies from CDC indicated that in 1997-98 and again in 1999-2000, fewer than half of persons from whom fluoroquinolone-resistant strains had been isolated had a history of foreign travel. In 1997 and 1998, 21 (57%) of 37 persons with fluoroquinolone resistant infections had not been out of the country in the 7 days preceding illness (11), and in the 1999-2000 case-control study, 37 (58%) of 64 persons with fluoroquinolone resistant infections had not been out of the country in the 7 days preceding illness (12). Thus, following licensure of fluoroquinolone agents for use in poultry, the proportion of Campylobacterstrains from humans in this country that were resistant increased, and in CDC studies, more than half of the resistant infections were acquired in the United States.
The sources of these infections acquired in the United States have been investigated in a multisite case-control study, reported by Kassenborg et al. (12). Case-control studies are an important and frequently used epidemiologic method for determining the risk factors and preceding exposures that are associated with an infection. In regard to the association between having a resistant infection and exposure to poultry, CDC again refers AHI to the analysis of Kassenborg et al. (12). This analysis shows that fluoroquinolone-resistant Campylobacter infections acquired in the United States, compared to healthy controls, were associated with eating poultry cooked at a commercial establishment (i.e. in restaurants, cafeterias and other such locations). The analysis reported by Cox cited by you (3) does not appear to make this comparison, as it presents no analysis of exposures of fluoroquinolone resistant infections compared to healthy controls, and thus does not bear directly on this issue.
The NARMS database relied upon by CDC for its findings is developed from a population that is comparable to the general population. The NARMS surveillance of susceptibility of human Campylobacter strains is conducted at FoodNet sites. FoodNet is a sentinel network that is producing stable and accurate estimates of the burden of specific foodborne diseases in the United States. The FoodNet population under surveillance has been compared with the U.S. population in general based on census data is similar to the population of the United States; this similarity is sufficient to make the data useful for understanding the epidemiology of foodborne illnesses in general (13). Persons withCampylobacter in NARMS resemble persons reported with Campylobacter infection in FoodNet in general, by age and sex distribution, and so appear to be drawn from the same general population. The sentinel site surveillance program FoodNet, as part of the CDC Emerging Infections Program, provides the best available information for monitoring Campylobacter infections, as Campylobacter infection is not a nationally notifiable disease (12).
As noted in our original response, the way that Campylobacter strains are selected for susceptibility testing in NARMS is different from the way that Salmonella or E. coli O157:H7 strains are selected. This is because at the time NARMS was begun, Salmonella and E. coli O157:H7 bacteria isolated from patients in clinical laboratories were routinely sent to State public health laboratories for further characterization, whileCampylobacter isolates were not routinely referred in this way, and so were not available for selection at the State laboratory in all locations. As a result the sampling framework for Campylobacter depends on a panel of participating clinical laboratories in FoodNet sites. The sampling and testing procedures used for Campylobacter surveillance through NARMS have been consistent since it began, so the changes observed are not attributable to changes in the surveillance method.
While CDC agrees that poultry is not the only source of Campylobacter infections, CDC and other investigations have shown that the dominant source of Campylobacter infections in the U.S. population is exposure to poultry. Other exposures, including consumption of other meats, unpasteurized milk, untreated surface water, and contact with animals have also been linked to this infections (14). CDC agrees that environmental pollution with antimicrobial agents in agricultural runoff or human waste is a potential concern for fluoroquinolone resistance in the future. However, the contribution of waterborne Campylobacter infections in the United States appears to be small at this time, as measured by case-control studies, and this exposure was not identified as a risk factor in the study focusing on fluoroquinolone-resistant infections (12). It has been shown that poultry, particularly chicken, is frequently contaminated with fluoroquinolone resistant Campylobacter(2,15).
Likewise, your reliance on outbreak data is not reliable in determining the sources of reported fluoroquinolone-resistant Campylobacterinfections. The reported outbreaks of Campylobacter have differed substantially in seasonality from the far more common sporadic cases, and it has been noted that in the United States the distribution of sources for outbreaks also differ from those identified for sporadic cases (14). Reported series of Campylobacter outbreaks suggest the range of potential sources of infections, but are not a definitive guide to the relative importance of the several sources of sporadic cases.
CDC agrees that statistical association of Campylobacter infection with an exposure such as eating undercooked poultry is not by itself a demonstration of causality. The epidemiologic judgment of causality depends on an array of associated observations bearing on the biologic and epidemiologic plausibility, and likelihood of a real causal association. The combination of statistical association and the matrix of related observations, (e.g. the repeated observation of the association in different studies, the high frequency of contamination of poultry withCampylobacter, the similarity of the strains in poultry and the strains in humans, and association with specific practices such as handling raw poultry) leads to determination of a causal association.
Thus, the resistance of Campylobacter jejuni isolated from humans in the United States to fluoroquinolones was 0% in 1989-90, and increased more recently to an appreciable fraction, 18%, in 2001 (10). The proportion of the resistant infections associated with foreign travel is less than half in recent studies, and consumption of poultry in commercial establishments is the dominant risk factor identified for resistant infections acquired in the United States. Therefore, the statement that proportion of Campylobacter that are resistant has increased and that it is associated with consumption of poultry outside the home is supported by our analysis.
CDC agrees that further analysis of the changes in incidence of resistant and susceptible infections would be useful, in addition to the analyses based on the proportion that are resistant.
It is possible that the incidence of susceptible infections is decreasing, while that of resistant infections is increasing.
CDC also notes that it will be difficult to interpret current and future trends in fluoroquinolone resistance in Campylobacter without information about the actual amount of fluoroquinolone agents given to poultry every year. If such use selects for resistant strains ofCampylobacter, then an increase in the amount used would be expected to lead to an increase in resistant Campylobacter contamination of poultry, and an increase in resistant infections in consumers of the poultry. Similarly, a decrease in use could lead to a decrease in observed resistant infections. Thus a future decrease in the incidence of resistant infections could be quite consistent with the association between use of fluoroquinolone agents in poultry and the appearance of resistant infections in humans, if the use of fluoroquinolones in poultry decreased. Regrettably, information on the amount used is not available.
REQUEST FOR CORRECTION 3: AHI requests that CDC correct its statement that estimate that there are approximately 2.4 million cases of campylobacteriosis in humans annually in the U.S. because these statements do not take into account the reduction in campylobacteriosis cases in recent years.
CDC Response to the appeal: For the reasons stated below, CDC denies this request as it applies to past reports, but agrees that it should be considered in the future. The estimate for the total burden of Campylobacter infection in the United States published in 1999 remains the most recently published estimate (16). The decline in Campylobacter infections since 1996 as monitored in FoodNet was first analyzed in a statistical model that provided a significance test in 2002 (17). This observation was repeated in 2003, after adding an additional year of surveillance data, which showed that between 1996 and 2002, the modeled incidence of Campylobacter infections decreased significantly by 24% (18). The continued surveillance data published in 2003 helps demonstrate that the decline in Campylobacter infections is sustained. We agree that when we use the 2.4 million estimate now, we should qualify it with a statement that the estimate was published in 1999, and by a statement that we are observing a recent decrease in incidence (presuming, of course that this decrease does continue).
Conclusion
In summary, upon careful review of your appeal, CDC does not find that the statements made in CDC abstracts or presentations to which you refer misrepresent the available data or require corrective action. However, we agree that presentations should clearly state what data are preliminary and that final results may differ from preliminary reports. CDC also agrees, that as time passes, the use of estimates of incidence published in 1999 should be noted as dating from that time, until such time as they are replaced by a new set of estimates. Future statements about trends should also note the difficulty of assessing them in the absence of reliable information about amounts of fluoroquinolone agents used in animal production.
We appreciate your comments and trust that this information helps clarify CDC's efforts to communicate to the public.
Sincerely,
/s/
Dixie E. Snider, Jr., M.D., M.P.H.
Assistant Surgeon General
Acting Deputy Director for Public Health Science
Enclosure
1. McClellan, J., et al. Prevalence and consequences of fluoroquinolone-resistant Campylobacter infections: NARMS 1997-2000. inInternational Conference on Emerging Infectious Diseases. 2002. Atlanta, GA: American Society for Microbiology.
2. Smith, K., et al., Quinolone-resistant Campylobacter jejuni infections in Minnesota, 1992-1998. New England Journal of Medicine, 1999.340: p. 1525-1532.
3. Cox, L., Re-examining the causes of campylobacteriosis. International Journal of Infectious Diseases, 2002. 6(Supplement 3): p. 3S26-3S36.
4. Coker, A., et al., Human campylobacteriosis in developing countries. Emerging
Infectious Diseases, 2002. 8(3): p. 237-243.
5. Tenover, F., et al., Antimicrobial resistance in Campylobacter species, in Campylobacter jejuni; Current status and future trends, I. Nachamkin, M. Blaser, and L. Tompkins, Editors. 1992, American Society for Microbiology: Washington D.C. p. 66-73.
6. Nachamkin, I., Antimicrobial susceptibility of Campylobacter jejuni and Campylobacter coli to ciprofloxacin, erythromycin and tetracycline from 1982 to 1992. Medical Microbiology Letters, 1994. 3: p. 300-305.
7. Kiehlbauch, J., M. Simon, and J. Makowski, Use of filtration to isolate Campylobacter and related organisms from stools, inCampylobacters, helicobacters and related organisms, D. Newell, J. Ketley, and R. Feldman, Editors. 1996, Plenum Press: New York. p. 47-49.
8. Endtz, H., et al., Quinolone resistnace in campylobacter isolated from man and poultry following the introduction of fluoroquinolones in veterinary medicine. Journal of Antimicrobial Chemotherapy, 1991. 27: p. 199-208.
9. Nachamkin, I., H. Ung, and M. Li, Increasing fluoroquinolone resistance in Campylobacter jejuni, Pennsylvania, 1982-2001. Emerging Infectious Diseases, 2002. 8: p. 1501-1503.
10. Anderson, A., et al. Fluoroquinolone-Resistant Campylobacter jejuni Infections in the United States: NARMS Data, 1997 - 2001. inNational Antimicrobial Resistance Monitoring Systems. Annual Scientific Meeting. November 19-22. 2002. Hilton Head, South Carolina.
11. Gupta, A., et al. Fluoroquinolone-resistant Campylobacter jejuni infections in the United States, 1997-2000: National antimicrobial resistance monitoring system's data lead to regulatory action. in Infectious Disease Society of America Annual Meeting. 2000. San Francisco, California.
12. Kassenborg, H., et al. Eating chicken or turkey outside the home associated with domestically acquired fluoroquinolone-resistant Campylobacter infecions: A FoodNet case-control study. in 2nd International Conference on Emerging Infectious Diseases. 2000. Atlanta, GA.
13. Hardnett, F., et al. Comparability of FoodNet and United States populations. in 3rd International Conference on Emerging Infectious Diseases. 2002. Atlanta, GA: American Society for Microbiology.
14. Friedman, C.R., et al., Epidemiology of Campylobacter jejuni infections in the United States and other industrialized nations, inCampylobacter, 2nd edition, M. Blaser and I. Nachamkin, Editors. 2000, American Society for Microbiology: Washington, D.C. p. 121 -138.
15. Rossiter, S., et al. Frequent isolation of fluoroquinolone-resistant Campylobacter from ill humans and chickens in grocery stores in the US. . in 137th American Veterinary Medical Association Annual Meeting. 2000. Salt Lake City, Utah.
16. Mead, P., et al., Food-related illness and death in the United States. Emerging Infectious Diseases, 1999. 5(5): p. 607-625.
17. CDC, Preliminary FoodNet data on the incidence of foodborne illnesses - selected sites, United States, 2001. Morbidity Mortality Weekly Report, 2002. 51: p. 325-29.
18. CDC, Preliminary FoodNet data on the incidence of foodborne illnesses - selected sites, United States, 2002. Morbidity Mortality Weekly Report, 2003. 52: p. 340-343.