Requestor: Kent McClure
Organizational Affiliation: Animal Health Institute
Date of request: 12/13/2002
Description of Information:
In 1994, the Director of the CDC ("Director") wrote to the Food and Drug Administration ("FDA") Director of the Center for Veterinary Medicine ("CVM") opposing the granting of approval of new animal drug applications for use of fluoroquinolones in livestock generally, and including specifically for use in poultry. (Exhibit 1.) Fluoroquinolones represented a newer class of antibiotics already approved for use in humans and pets, and the Director did not want their use to be extended to poultry.
The Director expressed concern that the use of fluoroquinolones in poultry could lead to the development of fluoroquinolone resistant food borne bacteria, such as Campylobacter, making it more difficult to treat with fluoroquinolones, people who develoaed campylobacteriosis as a result of eating undercooked poultry. To address concerns about licensure, such as those voiced by the Director, FDA convened a public hearing of its advisory committees for anti-infectives used in human medicine and the veterinary medicine advisory committee. After hearing from veterinary and human medical experts, FDA's committees jointly recommended to CVM that fluoroquinolones could be approved for use in poultry, subject to certain limitations. To minimize the potential for veterinary use of fluoroquinolones to lead to development of fluoroquinolone resistant bacteria that could affect public health, CVM adopted the recommendations of the advisory committees, including prohibiting use of fluoroquinolones, except to treat disease (i.e., no non-therapeutic use), requiring a prescription from a veterinarian prior to use, and prohibiting any off label use in food producing animals. CVM licensed the first fluoroquinolone for use in poultry in late 1995, and an additional fluoroquinolone in poultry in 1996.
Specific reason(s) why information does not comply with OMB, HHS or CDC guidelines:
Since 1995, and continuing to the present, certain personnel at CDC have undertaken an effort to convince FDA to withdraw the fluoroquinolone approvals for poultry. In furtherance of this effort, these personnel have,
1. used unpublished and/or preliminary data to present abstracts at scientific meetings, and to make other written and oral, scientific and media presentations, publications and statements, including those identified below, that are not otherwise supported by the data ("Statement-1 ", "Statement-2", "Statement-3", collectively "Statements");
2. used unpublished and/or preliminary data to support Statements, while delaying release of and/or otherwise resisting requests made for the data, for the purpose of peer review, including requests made pursuant to the Freedom of Information Act ("FOIA");
3. cited published literature to support Statements, which literature CDC personnel should have known and/or had reason to believe were incorrect;
4. used Statements to convince CVM to commence an action seeking to withdraw the approval of fluoroquinolones for use in poultry in the United States. Said action was commenced by CVM on October 31, 2000, and continues to the present. (Exhibit 2).
B. Statements
Statements include, but are not limited to the following:
1. Extended illness: A person with a Campylobacter infection that is resistant to fluoroquinolones will have an extended illness or two days of diarrhea more than a person who is infected with a Campylobacter that is susceptible to fluoroquinolones. ("Statement-l") In fact, the scientific data reasonably believed to be relied on by CDC personnel to support Statement-1 do not in fact support Statement-1, as further detailed below.
2. Rapid rise in resistance attributable to poultry: 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 fluoroquinolones in poultry. ("Statement-2".) In fact, the scientific data reasonably believed to be relied on by CDC personnel to support Statement-2, do not fairly support Statement-2, as further detailed below.
3. Occurrence of campylobacteriosis in U.S.: There are approximately 2.4 million cases of campylobacteriosis in humans annually in the United States. ("Statement3".) I:n fact, this estimate is inaccurate because it relies on older data and does not take into account the dramatic reduction in campylobacteriosis case over the at least five years, as expressly acknowledged on CDC's website, as further detailed below.
C. Basis of Conclusion that Statements are in Error
1. Statement-1: Extended Illness. Statement-1, and similar publications by CDC :personnel concerning the longer duration of illness caused by or associated with a fluoroquinolone resistant Campylobacter, are, to the best of Requester's knowledge, information and belief, significantly if not completely based on misinterpretations of the following studies/data.
A. CDC Dataset.3 The CDC Dataset is a CDC conducted case control study of campylobacteriosis, based on data collected during 1998-1999.
Statement-1 was reviewed by independent experts, based on an analysis of the raw data, questionnaire and other information comprising the CDC Dataset, obtained from CDC under FOIA. The independent experts concluded that the CDC Dataset, analyzed, according to generally accepted principles of epidemiology does not show any difference in duration of illness between resistant and susceptible Campylobacter. (Exhibit 3.) An extended duration of illness can be demonstrated only by including in the data analyzed individuals who had taken a fluoroquinolone, previous to being cultured for Campylobacter ("prior use") and/or those who had traveled to a foreign country before being infected with Campylobacter ("foreign travel"). In cases where the infection was acquired domestically and where there was no prior use there was no difference in duration of illness between resistant and susceptible organisms.
It is inappropriate scientifically to include foreign travel and prior use in an analysis to determine whether use of an antibiotic in poultry is the cause of resistant human infections in the United States. Whatever the cause of a resistance infection acquired outside: the U.S., it is not likely related to use of fluoroquinolones in poultry in the U.S. Additionally, foreign travel is a well-recognized confounding variable that must be controlled for in epidemiological studies of enteric disease by excluding persons who have recently traveled to a foreign country. Similarly, since treatment of campylobacteriosis with a fluoroquinolone can lead to fluoroquinolone-resistant Campylobacter one cannot tell whether resistant Campylobacter isolated after treatment was the result of the treatment or another cause, such as use of the antibiotic in poultry. Requester has been informed by independent experts, including former CDC personnel that the general practice in conducting similar case control studies would be to exclude foreign travel and prior use related cases in the analysis.
While Requester has been informed that a manuscript for the CDC Dataset was completed by the authors more than a year ago, to date no complete paper has been published in a peer review journal, in any other scientific publication, and no manuscript has been made available to the public. Notwithstanding the failure to publish a scientific paper CDC personnel have for several years, including since October 1, 2002 made statements to the media and otherwise publicizing Statement-1. CDC employees have resisted timely public access to the raw data and the CDC Dataset, which were finally obtained under FOIA, only after repeated requests and an extended period of time. (Exhibiit 4.)
B. Smith et al. CDC personnel have also sought to support Statement-1 by reliance on the conclusions from a published article by Smith et al. (Exhi bit 5.) Smith et al. conducted a case control study of risk factors for campy] obacteriosis in Minnesota, based on data collected in 1992-1998. Only by including foreign travel can Smith be fairly read to support Statement-1. Absent foreign travel independent expert analysis of Smith's raw data (obtained under the Minnesota FOIA) and related documentation demonstrate no additional days of illness contingent on whether the organism is resistant or susceptible to fluoroquinolones. (Exhibit 3.) To the best of Requester's knowledge, information and belief CDC personnel have never reviewed Smith's raw data. CDC personnel have been informed and otherwise should have known that Smith cannot fairly support Statement-1.
2. Statement-2: Rapid rise in resistance. Statement-2, and similar publications by CDC personnel are, to the best of Requester's knowledge, information and belief, significantly if not completely based on misinterpretations of the following studies/data.
A. Data from the National Antimicrobial Resistance Monitoring SystemS"NARMS ,).5 NARMS is a sentinel monitoring system, collecting from various laboratories, human and veterinary isolates of several bacteria, including Campylobacter, and testing the bacteria for resistance to certain anti-microbials, including ciprofloxacin, a fluoroquinolone. CDC has responsibility for resistance testing, maintenance and control of the NARMS database containing bacteria isolated from humans. USDA has responsibility for resistance testing, maintenance and control of the NARMS database containing bacteria isolated from animals.
Statement-2 was reviewed by independent experts, using standard and widely accepted statistical and epidemiological tools, based on an analysis of the NARMS protocol for collection of bacterial isolates, raw data from the NARMS collection (obtained from CDC under FOIA), and interviews with people involved in collection and testing of data, including CDC and USDA personnel. The independent experts concluded that NARIV[S was not designed to, does not, and cannot be used to compare fluoroquinoloneresistant Campylobacter from year-to-year because of many design, methodological and other flaws. (Exhibit 6.) Senior personnel at CDC involved in the NARMS program have publicly confirmed that NARMS is not representative of the national prevalence of fluroquinolone resistant Campylobacter and that NARMS cannot be used for year to year comparisons of national resistance rates. (Exhibit 7.) For example, on November 20, 2002, in response to a presentation made at the 2002 NARMS Annual Scientific Meeting ("2002 NARMS Meeting") CDC personnel acknowledged that the NARMS data were not a representative sample of the national prevalence of resistant Campylobacter.
The NARMS protocol requires state reporting laboratories to send one Campylobacter isolate weekly to CDC. CDC does resistance testing and the results are published on a yearly basis. However, CDC personnel have made numerous oral presentations and scientific abstracts have been presented at various scientific meetings by such personnel, including Statement-2, based on partial year, preliminary and or otherwise incomplete data. Since the first two quarters of the year generally show higher fluoroquinolone resistant rates than the full year, publication of early or preliminary data tend to overstate the true yearly resistant rate as determined by NARMS. CDC personnel are aware that the early quarters overstate the annual resistance rates.
For example, a CDC publication from April 1998 purports to show, as part of a NARMS update, fluoroquinolone resistance in Campylobacter in 1997 of 13.2%, thus far in 1998 resistance of 15.6%6 and that the data indicate the emergence and widespread dissemination of fluoroquinolone resistant Campylobacter in the United States sometime between 1990 and 1997. (emphasis added) (Exhibit 8.) In fact, the NARMS fluoroq uinolone resistance rate for 1998 was 13.1%.
In addition, a March 2002 presentation by CDC personnel entitled "Human health consequences of antimicrobial use in agriculture" (Exhibit 9), the author presented early and incomplete data from NARMS 2001, showing a slide and stating at the meeting that the date. for Campylobacter resistance to ciprofloxacin in humans was at 25%.7 Neither the slide nor the presenter explained that the rate for the full year was expected to be lower. Only after an attendee at the meeting brought this to his attention, did the presenter acknowledge in private that the slide was inappropriate. However, he did not correct his remarks before the attendees. In fact, the preliminary resistance rate of 18% for the full year 2001, presented by CDC at the 2002 NARMS Meeting, is significantly lower than the partial year 25% rate presented at the March 2002 meeting.
Further, on November 20, 2002 CDC presented a poster ("Poster") at the 2002 NARMS Meeting which represented that 19% of the Campylobacter isolates tested for resistance in NARMS for 2001 were resistant to fluoroquinolones, notwithstanding that these data are preliminary. (Exhibit 10.) In response to questioning from conference attendees, CDC personnel acknowledged that the NARMS 2001 data on the poster were preliminary and would not be final for several months. An abstract contained in the printed materials for conference attendees represented the same 2001 NARMS Campylobacter resistance rate as 18%, but did in fact identify the 2001 data as preliminary. (Exhibit 11.)
CDC personnel have resisted and delayed legitimate efforts to timely and independent peer review the NARMS results, including attempts made under FOIA to obtain the raw data purporting to support published early, preliminary and yearly NARMS results. (Exhibit 12.)
However, groups supporting CDC's effort to ban fluoroquinolone use in poultry have been provided with preferential access to underlying information. For example, to the best of Requesters knowledge, information, and belief CDC personnel provided the group Keep Antibiotics Working ("KAW") with a copy of the Poster prior to release of the poster at the 2002 NARMS meeting. (Exhibit 10.) The Poster did not state that the data were preliminary nor did it disclose any other qualification on the interpretation of the numbers of isolates or percentage of isolates represented as resistant to fluoroquinolones. For example, the Poster does not disclose the fact that the percent resistance does not represent National prevalence. KAW presented these data, --including the 19% fluoroquinolone resistant Campylobacter rate for 2001 -- at a press conference on November 19, 2002. The Poster, now available on KAW's website,9 does not indicate that the 2001 NARMS findings are preliminary. (Exhibit 10.) CDC's representative presented the Poster at the 2002 NARMS Meeting, but would not provide copies of the Poster at that time to conference attendees. As of this date the Poster is still not available on CDC's website.
B. CDC Sentinel County Study 1989-90. CDC personnel have also sought to support their conclusion about rising rates of fluoroquinolone resistant Campylobacter by stating that there was little to no fluoroquinolone resistance Campylobacter prior to the introduction of fluoroquinolones in poultry. For example, on November 20, 2002, a published article quoted CDC personnel as saying that there was no Campylobacter resistance to ciprofloxacin in 1990 but that resistance has increased every year since. (Exhibit 13.) Such statements have been made based in part on reliance on a CDC Sentinel County study conducted on Campylobacter isolates collected during 19891990. Sobel et al. first published an abstract related to this study for an EIS conference in 1996.0 (Exhibit 14.) Subsequent CDC publications have cited this abstract to support the proposition that no resistant Campylobacter was found in the 1989-90 study, even though the abstract does not discuss resistance. (Exhibit 15.) A subsequent analysis attributed to Sobel et al. was published in April 1998 as part of a NARMS update, purporting to show fluoroquinolone resistance in Campylobacter in 1997 of 13.2%, thus far in 1998 resistance of 15.6% and concluded that the data indicate the emergence and widespread dissemination of fluoroquinolone resistant Campylobacter in the United States sometime between 1990 and 1997. (Exhibit 8.)
To the best of Requester's knowledge, information and belief the CDC Sentinel County study and the Sobel et al. analysis of this data do not conclude that there was no fluoroquinolone-resistant Campylobacter prior to the introduction of fluoroquinolones in poultry. However, CDC has refused inappropriately to provide the complete data used by Sobel et al. for peer review. (Exhibit 16.)
C. Smith et al. CDC personnel have also relied on Smith et al. to support: their conclusion about rising rates of fluoroquinolone-resistant Campylobacter. To the best of Requester's knowledge, information, and belief, CDC personnel have never reviewed Smith's raw data. Independent expert analysis of the raw data supporting Smith et al. demonstrates that small temporal trends such as those demonstrated in Smith et al. are unlikely to be interpretable due to under-reporting of cases, retrospective data collection, and recall bias. (Exhibit 3).
3. Statement-3: Occurrence of campylobacteriosis in U.S. In 1999 CDC published an estimate of food born illness in the United States which estimated 2.4 million campylobacteriosis cases in the United States annually ("Estimate"). The Estimate frequently used by CDC personnel is derived from a combination of outdated data and figures extrapolated based upon the etiology of Salmonellosis.13 Much of these data are in excess of ten years old. The Estimate exaggerates the number of annual cases of campylobacteriosis, because, in actuality, and according to figures published on CDC's website the rate of Campylobacter infection has decreased markedly in the past at least five years. CDC Food Net reported a 19% decline in the incidence of Campylobacter from 1998 to 1999 in the original five Food Net sites. 14 From 1996 to 2000 CDC reported a 27% drop in human cases of campylobacterosis.15 Independent analysis of the data indicates a 41% drop in human cases of Campylobacter. Recently, KAW used a rate of two million cases annually.' 6 Regardless of the actual percent reduction in the annual incidence of campylobacteriosis in the U.S., the level of human Campylobacter infection has decreased significantly between 1996 and 2001 and statements by CDC personnel therefore overstates significantly the annual number of Campylobacter cases.
Furthermore, the extrapolation of the data used by Mead assumed: 1) a consistent rate of campylobacteriosis throughout the entire United States population and 2) that the extrapolation rates for Salmonellosis apply to campylobacteriosis. These assumptions further call into question the validity of this figure.
II. Places Where Statements Are Being Disseminated
Appendix A lists the places, among others where Statements have been, and in most cases continue to be, disseminated. Since the CDC Information Quality Guidelines became effective on October 1, 2002, at least eleven presentations and abstracts have been disseminated on the CDC website that make erroneous statements containing factual inaccuracies not supported by a fair reading of the underlying data. CDC personnel presentations on November 20, 2002 continue this pattern and practice.
Requestor's recommendation for correction:
Requester recommends that the following steps be taken to insure that the information listed above complies with the standards for information quality outlined in the CDC Guidelines.
- The existing and similar documents should be modified, either by removing the inaccurate statements or prominently indicating corrections to these statements. In addition, CDC should mandate that future statements, speeches, abstracts, articles, and any other form of dissemination by the Agency will comply with the results of this corrective challenge.
- CDC should in the future refrain from making statements about rates, causes or other statements about antibiotic resistance based on partial or preliminary data when it is reasonably likely that the results from the final or complete data will differ significantly from preliminary or partial data.
- CDC should make data it uses to support public statements concerning rates, causes, or other factors concerning antibiotic resistance readily available to the public for peer review.
- CDC should refrain from making any statements interpreting NARMS without qualifying the limitations of the data.
- CDC should issue a corrective statement, clearly indicating the specific information that has been found to be in error pursuant to this corrective challenge, and should prominently display the corrective statement on its website and in other appropriate venues of dissemination. The corrective statement should also be attached to every archival document that references the information that has been corrected.
How the requestor was affected by the information?
- Erroneous information hares human and animal health by producing data incorrectly attributing human health affects to animal drugs. To the extent that this erroneous information results in unnecessary reductions in the use of beneficial animal drugs, the overall health of both humans and animals will be negatively affected. A fluoroquinolone product is approved for, effective in, and essential to treat life threatening bacterial infections in poultry. It is the only practicable alternative for reducing bacterial infections in poultry. Reductions in poultry bacterial infections result in reductions in human bacterial infections.
- The use of flawed information and unsupported reports sets a bad precedent, generally adversely effecting scientific debate, setting of public policy, and regulatory priorities.
- Information errors undermine the credibility of the CDC.
- The use of flawed information and unsupported reports sets a bad precedent, generally adversely effecting any drug manufacturer or association involved in efforts to improve animal health.
- Reports in the media based on inaccurate information harp the poultry industry through the creation of an erroneous conclusion that poultry may be unsafe.
Last Revised: August, 2004