Document RaapxxOKYEKdXKNjrXO3nOZ47

FILE NAME: Doubt Science (DBTS) DATE: 2006 DOC#: DBTS021 DOCUMENT DESCRIPTION: Journal Article from NY Academy of Sciences Manufactured Uncertainty: Protecting Public Health in the Age of Contested Science and Product Defense Manufactured Uncertainty Protecting Public Health in the Age of Contested Science and Product Defense DAVID MICHAELS Department o f Environmental and Occupational Health, The George Washington University School o f Public Health and Health Services, Washington DC 20037, USA Abstract: The strategy of "manufacturing uncertainty" has been used with great success by polluters and manufacturers of dangerous prod ucts to oppose public health and environmental regulation. This strat egy entails questioning the validity of scientific evidence on which the regulation is based. While this approach is most identified with the tobacco industry, it has been used by producers of asbestos, benzene, beryllium, chromium, diesel exhaust, lead, plastics, and other hazardous products to avoid environmental and occupational health regulation. It is also central to the debate on global warming. The approach is now so common that it is unusual for the science not to be challenged by an industry facing regulation. Manufacturing uncertainty has become a business in itself; numerous technical consulting firms provide a ser vice often called "product defense" or "litigation support." As these names imply, the usual objective of these activities is not to generate knowledge to protect public health but to protect a corporation whose products are alleged to have toxic properties. Evidence in the scientific literature of the funding effect--the close correlation between the results of a study desired by a study's funder and the reported results of that study--suggests that the financial interest of a study's sponsors should be taken into account when considering the study's findings. Similarly, the interpretation of data by scientists with financial conflicts should be seen in this light. Manufacturing uncertainty is antithetical to the public health principle that decisions be made using the best evidence currently available. . Keywords: uncertainty; certainty; doubt; regulation; product defense; litigation support; funding effect; public health Address for correspondence. David Michaels, Ph.D., M.P.H., Department of Environmental and Occupational Health, The George Washington University School of Public Health and Health Services, 2100 M St. NW, Suite 203, Washington DC, 20037. Voice: 202-994-2461; fax: 202-994-0011. e-mail: eohdinm@gwumc.edu Ann. N.Y. Acad. Sei. 1076: 149-162 (2006). 2006 New York Academy of Sciences, doi: 10.1196/annals.1371.058 149 150 ANNALS NEW YORK ACADEMY OF SCIENCES INTRODUCTION In March 2002, a nuclear reactor near Toledo, OH, came within a quarter inch o f a major radiation release, possibly the worst accident o f this type in U.S. history. Water mixed with boric acid had eaten through six inches o f carbon steel, leaving only a thin layer o f stainless steel to contain the water in the D avis-Besse nuclear reactor's vessel head. W hen finally seen by safety inspectors, that last steel layer was bulging, barely able to contain the highly pressurized coolant. Three months earlier, two other reactors had developed similar cracks. After studying the situation, experts at the U.S. Nuclear Regulatory Commission (NRC) predicted a high probability o f finding cooling system breaches at the Ohio plant and asked operators o f all similar reactors to shut down voluntarily and inspect for damage. The operator of Davis-Besse refused and NRC staff prepared an order demanding that -the reactor be shut down and inspected. But that order was never issued. Desiring to protect the financial health o f the operator, the NRC manager demanded "absolute p ro o f' that the vessel head was damaged before he would order a shut down and inspection, proof that could only be obtained with the shut down and inspection.1 Absolute certainty in the realm o f medicine and public health is rare. Our public health programs will not be effective if absolute p ro o f is required before we act; the best available evidence must be sufficient. Yet we see a growing trend that demands p ro o f over precaution in the realm o f public health.2 Few scientific challenges are more complex than understanding the cause of disease in humans. Scientists cannot feed toxic chemicals to people to see what dose causes cancer. Instead, we must harness the "natural experiments" where exposures have already happened in the field. In the laboratory, we can use only animals. Both epidemiologic and laboratory studies have many uncertainties, and scientists must extrapolate from study-specific evidence to make causal inferences and recommend protective measures. Absolute certainty is rarely an option. Our regulatory programs will not be effective if such proof is required before we act; the best available evidence must be sufficient. THE TOBACCO ROAD Years ago, a tobacco executive unwisely committed to paper the perfect slogan for his industry's disinformation campaign: "Doubt is our product. "3 With tobacco, doubt turned out to be less addictive for the public than the leaf itself, and the industry finally abandoned its strategy. I call this strategy "manufacturing uncertainty,"4 and no industry manu factured more uncertainty over a longer period than the tobacco companies. Following a strategic plan developed in the mid 1950s by the public relations MICHAELS: MANUFACTURED UNCERTAINTY 151 firm Hill and Knowlton, a firm that manufactured uncertainty on behalf of various industries over the course o f decades, Big Tobacco hired scientists to challenge the growing consensus linking cigarette smoking with lung cancer and other adverse health effects. This industry campaign had three basic mes sages: Cause and effect relationships have not been established in any way; statistical data do not provide the answers; more research is needed. As re cently as 1989, a spokeswoman appearing on national television dismissed claims that tobacco caused lung cancer as . .just statistics. The causal rela tionship between smoking and cancer has not yet been established."5 The industry even started its own "scientific" publication, Tobacco and Health Research, for which the main criterion for articles was straightforward: " .. .the most important type o f story is that which casts doubt on the cause and effect theory o f disease and smoking." Editorial guidelines stated that headlines "should strongly call out the point-- Controversy! Contradiction! Other Factors! Unknowns!"6 ` Doubt turned out to be less addictive for the public than tobacco, and the industry finally abandoned its strategy. Thanks to its efforts, however, public health protections and compensation for tobacco's victims were delayed for decades. The practices perfected by tobacco executives and public relations are alive and well. Learning from tobacco, other industries have discovered that debating the science is much easier and more effective than debating the policy. Witness the debate over global warming. Many studies link human activity, especially burning o f carbon fuels, with global warming.7 Waiting for absolute certainty that the accumulation o f greenhouse gases will result in dramatic changes in the climate seems far riskier and potentially far more expensive to address than acting now to control the causes of global warming. Opponents o f preventive action, led by the fossil fuels industry, avoid this policy debate by challenging the science instead with a classic uncertainty campaign. I need only cite a memo from the political consultant Frank Luntz, delivered to his clients in early 2003. In "Winning the Global Warming Debate," Luntz wrote: Voters believe that there is no consensus about global warming within the scientific community. Should the public come to believe that the scientific issues are settled, their views about global warming will change accordingly. Therefore, you need to continue to make the lack o f scientific certainty a primary issue in the debate... The scientific debate is closing [against us] but not yet closed. There is still a window o f opportunity to challenge the science, (emphasis in original)8 There has been substantial media coverage of the political machinations behind the global warming debate, and we all know about the behavior o f the tobacco industry. Less well known are the campaigns mounted to question studies documenting the adverse health effects o f exposure to beryllium, lead, mercury, vinyl chloride, chromium, benzene, benzidine, nickel, and a long list 152 ANNALS NEW YORK ACADEMY OF SCIENCES o f other toxic chemicals and pharmaceuticals. In fact, it is unusual for the science behind any proposed public health or environmental regulation not to be challenged, no matter how powerful the evidence. How ridiculous can it get? There is widespread agreement in the scien tific community that broad-spectrum ultraviolet (UV) radiation, whether from sunlight or from tanning lamps, causes skin cancer. Yet trade associations representing the indoor tanning industry have attempted to derail the "cancercausing" designation by questioning the scientific evidence.9 Manufacturing uncertainty on behalf of big business has become a big busi ness in itself. The "product defense" firms have become experienced, adept, ' and successful consultants in epidemiology, biostatistics, and toxicology. The work o f these product defense firms bears the same relationship to science as the Arthur Andersen Company does to accounting-- or did before it went bankrupt following the Enron debacle. BERYLLIUM: NATIONAL DEFENSE OR "PRODUCT DEFENSE"? The metal beryllium is extremely useful and almost unimaginably toxic. Breathing the tiniest amount of this lightweight metal can cause disease and death. As a neutron moderator that increases the yield o f nuclear explosions, beryllium is vital to the production o f weapon systems. Throughout the cold war, the U.S. nuclear weapons complex was the nation's largest consumer o f the substance. As a result, however, hundreds o f weapons workers-have developed chronic beryllium disease (CBD). It is not ju st machinists who work directly with the metal who develop CBD, but also others simply in the vicinity o f the milling and grinding processes, often for very short periods o f time, and even people living near beryllium factories. As Assistant Secretary o f Energy for Environment, Safety, and Health from 1998 to 2 0 0 1 ,1 was the chief safety officer for the nuclear weapons complex, responsible for protecting the health o f workers, the communities, and the envi ronment around the production and research facilities. In 1998 the Department o f Energy's (DOE) exposure standard had been unchanged for almost 50 years, and there were hundreds o f cases o f beryllium disease in the nuclear weapons complex and in factories that supplied beryllium products. The history o f this original DOE beryllium standard is legendary. It was developed in a 1948 discussion held in the back seat of a taxi by Merril Eisenbud, an Atomic Energy Commission (AEC) industrial hygienist, and Willard Machle, a physician who was a consultant to the firm building the Brookhaven Laboratory in Long Island, New York. Eisenbud discusses this history in his autobiography, noting that they selected the exposure limit "in the absence o f an epidemiological basis for establishing a standard" 10 (p. 55). The AEC "tentatively" adopted a standard o f 2 ug/m3 in 1949, and then re viewed it annually for 7 years before permanently accepting it. MICHAELS: MANUFACTURED UNCERTAINTY 153 When first implemented, the 2 ug/m3 standard resulted in a dramatic de crease in new beryllium disease cases. But by 1951, Eisenbud recognized that the the distribution of the chronic form o f beryllium disease did not follow the usual exposure-response model seen for most toxic substances and hy pothesized an immunological susceptibility.11 It was not long before CBD was seen among workers hired after the 1949 standard went into effect, and whose exposure appeared to be below the 2 ug/m 3 standard.12 Moreover, CBD had been diagnosed in persons with no workplace exposure to the metal, including individuals who simply laundered the clothes o f workers, drove a milk delivery truck with a route near a beryllium plant, or tended cemetery graves near a beryllium factory.12 When the Occupational Safety and Health Administration (OSHA) was established in 1971 to protect the health o f workers in the private sector, it simply adopted the taxicab standard. By the 1980s, however, it was clear that workers exposed to beryllium levels well below the standard were develop ing disease. As both the DOE and OSHA began the time-consuming legal process o f changing their standards, the beryllium industry objected. At one public meeting, the Director of Environmental Health and Safety of Brush Wellman, the leading U.S. producer o f beryllium products asserted (according to D O E's minutes o f the meeting): "Brush Wellman is unaware o f any scien tific evidence that the standard is not protective. However, we do recognize that there have been sporadic reports o f disease at less than 2 ug/m 3. Brush Wellman has studied each o f these reports and found them to be scientifically unsound." 13 In 1991, Brush managers were told that if they were "asked in some fashion whether or not the 2 ug/m3 standard is still considered by the company to be reliable," they should answer "In most cases involving our employees, we can point to circumstances o f exposure (usually accidental), higher than the standard allows. In some cases, we have been unable (for lack o f clear history) to identity such circumstances. However, in these cases we also cannot say that there was not excessive exposure." 14 This was the industry's prim ary argument, and it was based on a flawed logic. It was not difficult to go back into the work history o f anyone with CBD and estimate that at some point in time, the airborne beryllium level must have exceeded the standard. Brush did this and then reasoned that the 2 ug/m3 must be fully protective since most people who had CBD had, at some point, been exposed to levels above the standard. The ever-increasing number of CBD cases identified at facilities across the nuclear weapons complex as well as in the beryllium industry's own factories rendered less plausible the claim that the old standard was safe. In Septem ber 1999 Brush Wellman sponsored a conference, in collaboration with the American Conference o f Governmental Industrial Hygienists, to bring "lead ing scientists together to present and discuss the current information and new research on the hazards posed by beryllium " 15 (p. 527). The papers were sub- 154 ANNALS NEW YORK ACADEMY OF SCIENCES sequently published together in an industrial hygiene journal. Clearly, one purpose o f the conference was to influence the government standard setting on beryllium: at the time o f the conference, DOE was a few months away from issuing its final rule and OSHA had signaled its intention to revise its outdated standard. . Several papers were presented by scientists employed by Exponent, Inc., the beryllium industry's product defense consultant. These included a paper entitled "Identifying an Appropriate Occupational Exposure Limit (OEL) for Beryllium: Data Gaps and Current Research Initiatives" that promoted the industry's new rationale for opposing a new, stronger beryllium standard: that ' more research is needed on the effects o f particle size, o f exposure to beryllium compounds and o f skin exposure to CBD risk. The paper concluded: "At this time, it is difficult to identify a single new TLV [threshold limit value] for all forms o f beryllium that will protect nearly all workers. It is likely that within three or four years, a series o f TLVs might need to be considered___ In short, the beryllium OEL could easily be among the most complex yet established" 15 (p. 536). After reviewing the public comments and the literature on beryllium 's health effects, the DOE health and safety office concluded that, while more research is always desirable, we had more than enough information to warrant immediate implementation of a stronger beryllium disease prevention standard. Over the industry's objections, we issued a new rule, reducing the acceptable workplace exposure level by a factor o f 10. Simultaneously, OSHA also recognized the inadequacy o f its own standard16 and announced its commitment to issuing a stronger one.17 However, when the George W. Bush Administration took office in 2001, the commitment to strengthening its beryllium rule was dropped from the agency's formal regulatory agenda. In November 2002 OSHA implicitly accepted the industry's approach by issuing a call for additional data on the relationship o f beryllium disease to, among other things, particle size, particle surface area, particle number, and skin contact.18 In the few years since DOE issued its standard, however, re searchers have published several epidemiologic studies that demonstrate that the 2.0 fxg/m3 standard does not prevent the occurrence o f CBD.19 22 In addition to CBD, the scientific community widely recognizes that beryl lium also increases the risk o f lung cancer23,24; several studies conducted by epidemiologists at the Center for Disease Control (CDC) support this conclu sion.2527 In 2002, however, scientists at a product defense firm published a 10-year-old reanalysis o f one o f the CDC studies.28 By changing some param eters, the statistically significant elevation o f lung cancer rates was no longer statistically significant. (Such alchemy is rather easily accomplished, of course, while the opposite-- turning insignificance into significance, is extremely dif ficult.) Not coincidentally, this particular firm had done extensive work for the tobacco industry.2 The new analysis was published in a peer-reviewed journal-- not one with much experience in epidemiology, but peer-reviewed MICHAELS: MANUFACTURED UNCERTAINTY 155 nevertheless, and the industry now touts its study as evidence that everyone else is wrong. And so it goes today, in industry after industry, with study after study, year after year. Data are disputed, data have to be reanalyzed. Animal data are deemed not relevant, human data not'representative, exposure data not reliable. More research is always needed. Uncertainty is manufactured. Its purpose is always the same: shielding corporate interests from the inconvenience and economic consequences of public health protections. PPA: THE TRICKS OF THE TRADE In order to attract new clients, some of these firms even brag about their successes. Until I wrote about it in Scientific American,2 the Weinberg Group (another firm that had worked extensively for the tobacco industry) advertised on its web site its contribution to the effort to oppose the Food and Drug Administration's (FDA) belated clampdown on phenylpropanolamine (PPA), the over-the-counter drug that was widely used for decades as a decongestant and appetite suppressant. Here is a short version o f the history o f PPA. Reports o f hemorrhagic strokes in young women who had taken a PPA-containing drug began circulating in the early 1970s. Twenty years later, when the FDA finally raised official questions about the safety o f PPA, the manufacturers rejected them. Eventually, a com promise was reached. The drug manufacturers would select an investigator and fund an epidemiologic study whose design would be jointly approved by the FDA. They chose the Yale University School o f Medicine. In October 1999 the manufacturers and the FDA learned that the study confirmed the causal relationship between PPA and hemorrhagic stroke.29 The study was published the following year in the New England Journal o fM edicine?0 W hen they initially learned o f the study's findings, did the manufacturers immediately withdraw this drug, which by then had annual sales of more than $500 million, but was responsible, according to an FDA analysis, o f between 200 and 500 strokes per year among 18-to-49-year-olds?31 No. Instead, they turned to the Weinberg Group to attack the study itself, focusing on "bias and areas o f concern."32 The manufacturers recognized that the FDA would eventually force the drug off the market, but they stalled for almost a year, enough time to reformulate their products. And when the FDA finally requested .manufacturers to stop marketing PPA in November 2000, the industry was prepared to ship reformulated products immediately.29 Here is the full text o f the web page on the Weinberg G roup's work on PPA: ADVERSE EVENT LINKED TO OTC PRODUCT A pharmaceutical company retained THE WEINBERG GROUP to audit the results of a FDA-requested, industry-sponsored case-control study that linked 156 ANNALS NEW YORK ACADEMY OF SCIENCES their over-the-counter (OTC) product and several others with a serious, lifethreatening adverse event. There was a substantial concern from the FDA based on reports of adverse events that use of these OTC products would present a public health problem. The study was commissioned to answer the question of risk with a controlled investigation. According to the study . investigators, the results of the study showed a strong association between these products and a severe, life-threatening adverse event. Epidemiologists at TFIE WEINBERG GROUP led experts and consultants to some of the other affected OTC companies, in an effort that included a reanalysis of the raw data from the case-control study, and an assessment of the study's methodological flaws. The unique ability of the experts at THE WEINBERG GROUP to combine their expertise in epidemiology and biostatistics with strategic thinking enabled them to lead the pharmaceutical company's effort in their dispute with the FDA.33 THE FUNDING EFFECT The biomedical literature extensively discusses the "funding effect," a term used to describe the close correlation between the results of a study desired by a study's funder and the reported results o f that study.34-36 Recent reviews in leading biomedical journals found that industry sponsorship was strongly associated with proindustry conclusions.37,38 The funding effect has also been seen in studies that look at the toxic effects o f chemical exposures. The disparity between the results of studies examining the.risk o f lung cancer among beryllium-exposed workers discussed above is an example of the funding effect: Three government-funded analyses find an elevated risk while the one industry-funded analysis (actually reanalysis) does not. An even more striking example in the toxicology literature is the debate over the effects o f low-dose exposure to bisphenol A (BPA), an environmental estrogen used in the manufacture o f polycarbonate plastic, a resin widely used in food cans and dental sealants. Exposure to BPA had been found in some studies to alter endocrine function at .very low doses. In response, the American Plastics Council hired the Harvard Center for Risk Analysis (HCRA) to conduct a weight-of-the-evidence review o f the toxicology. The HCRA panel reviewed 19 animal studies and reported that it found no consistent affirmative evidence o f low-dose BPA effects.39 This conclusion was challenged by scientists who felt that the HCRA had chosen to examine only a minority of the 47 studies available at the time. These scientists reviewed the 115 published that had been published through December 2004 and found results that differed markedly with the HCRA analysis.40 ` As can be seen in Table 1, 90% (94 of 104) of the studies paid for with government funds reported an effect associated with BPA exposure; not a MICHAELS: MANUFACTURED UNCERTAINTY 157 TABLE 1. Biased outcome due to source of funding in low-dose in vivo BPA research as of December 2004 Source of funding Government Chemical corporations Total Harm 94 0 94 Number of studies & effect reported No harm 10 11 22 Adapted from: Vom Saal, & Hughes.40 single one o f the 11 corporate funded studies found an effect. The correlation between sponsor and result requires no test of statistical significance beyond Joseph Berkson's test o f "interocular traumatic impact"-- the results hit you right between the eyes. VIOXX: CONFLICTED SCIENCE AND ITS CONSEQUENCES I am not presuming here that the scientists involved in "manufacturing un certainty" knowingly promote deadly products. More likely, scientists, along with the corporate executives and attorneys who hire them, convince them selves that the products they are defending are safe and that the evidence of harm is inaccurate, misleading, or trivial. This can be seen in the recent evidence on the cardiac effects o f Vioxx (rofecoxib), M erck & Co., Inc.'s blockbuster pain reliever that was taken off the market in November 2004, accompanied by headlines around the world. Even before the FDA approved Vioxx in May 1999, the agency reviewed data that suggested Vioxx could increase heart disease risk. Several independent scientists (i.e., not on M erck's payroll) also raised red flags, but for the most part, they were ignored by the FDA. Then the results o f a clinical trial appeared in early 2000, just a few months after the drug was put on the market, linking Vioxx with an increased risk o f heart attack. Merck had chosen naproxen (sold under the brand name Aleve) as the comparison treatment in the trial because aspirin, perhaps a more obvious choice, was known to lower cardiovascular .disease risk, and the company did not want its trial to show more heart attacks among the study participants who took Vioxx. But the results showed that participants who took Vioxx for more than 18 months had five times the risk o f heart attack as those taking naproxen.41 M erck's scientists faced a dilemma. They could interpret this finding to mean either that Vioxx increased heart attack risk by 400% or that naproxen was, like aspirin, beneficial in reducing the risk of heart attack by an astounding 80%. When a double-blind trial using a placebo control found seven excess heart 158 ANNALS NEW YORK ACADEMY OF SCIENCES attacks per every 1000 users per year, the correct interpretation was clear: Vioxx causes heart attacks. One FDA analysis estimates that Vioxx caused between 88,000 and 139,000 heart attacks, 30-40% o f which were fatal, in the 5 years the drug was on the market.42 Subsequent litigation has uncovered memos documenting that Merck exec utives were concerned about the increased risk o f heart attacks associated with Vioxx, but that they downplayed these concerns in their communications with physicians and resisted the FDA's efforts to add warnings to Vioxx's label.43 It is hard to imagine that the drug m aker's scientists were consciously promoting a product they knew would result in disease and death. At the same time, it is hard to imagine they honestly thought naproxen reduced the risk o f heart attack by 80% . It seems more likely that their allegiances were so tightly linked with the products they have worked on, as well as the financial health of their employers, that their judgm ent became fatally impaired. A NEW REGULATORY PARADIGM The lessons o f the past 40 or 50 years and the import o f the government's actions over the past 4 years are clear. A new regulatory paradigm is needed. Federal agencies must ensure that data and scientific analyses provided by manufacturers are independently verified. Opinions submitted to regulatory agencies by corporate scientists and, especially, the product defense industry must be taken as advocacy primarily, not as science. Below are a few steps that begin to approach this new paradigm. It has become apparent that some industry-supported research is never published because the sponsor did not like the results. Following a series o f alarming instances in which the sponsors of research used their financial control to the detriment o f the public's health, a group o f leading biomedical journals established policies that make their published articles transparent to commercial bias and that require authors to accept lull control and responsibility for their work. These journals will now only publish studies done under contracts in which the investigators had the right to publish the findings without the consent or control o f the sponsor. In a joint statement, the editors of the journals asserted that contractual arrangements allowing sponsor control of publication "erode the fabric o f intellectual inquiry that has fostered so much high-quality clinical research"44 (p. 1233). But the federal regulatory agencies charged with protecting our health and environment have no similar requirements. When studies are submitted to the EPA or OSHA, for example, the agencies do not have the authority to inquire who paid for the studies or whether these studies would have seen the light o f day if the sponsor did not approve the results. MICHAELS: MANUFACTURED UNCERTAINTY 159 Federal agencies should adopt, at a minimum, requirements for "research integrity" comparable to those used by biomedical journals: Parties submitting data from research they have sponsored must disclose if the investigators had the right to publish their findings without the consent or influence of the sponsor.45 It is also important to recognize that the opinions o f virtually any scientist can be clouded by conflict o f interest, even if it is not apparent to the scientist. Conflict of interest inevitably shapes judgment, and this must be factored into the consideration of the analyses and opinions o f scientists employed by industry. Public health is not well served by the unequal treatment of public and private science. While raw data from government-funded studies are generally available to private parties for inspection and reanalysis, enabling product defense experts to conductp o st hoc analyses that challenge troubling findings, .industry is under no obligation to release comparable raw data from their own studies. When private sponsors conduct research to influence public regulatory proceedings, these studies should be subject to the same access and reporting provisions as those applied to publicly funded science.46 Apologists for polluters and manufacturers of dangerous products commonly complain about government regulation, asserting that the agencies are not using "sound science." In fact, many o f these manufacturers o f uncertainty do not want "sound science"; they want something that sounds like science but lets them do exactly what they want to do. We all recognize that the science is just one part o f policy making. In shaping rules and programs to protect the public health and environment, decision mak ers also have to consider economic issues, values,' and a host o f other factors. In our current regulatory system, debate over science has become a substitute for debate over policy and the values upon which policy should be based. Opponents o f regulation use the existence o funcertainty, no matter its magni tude or importance, as a tool to counter imposition o f public health protections that may cause them financial difficulty. It is important that those charged with protecting the public's health recognize that the desire for absolute scientific certainty is both counterproductive and futile. This recognition underlies the wise words o f Sir Austin Bradford Hill delivered in an address to the Royal Society o f Medicine in 1965: All scientific work is incomplete--whether it be observational or experimen tal. All scientific work is liable to be upset or modified by advancing knowl edge. That does not confer upon us a freedom to ignore the knowledge we already have, or to postpone action that it appears to demand at a given time. Who knows, asked Robert Browning, but the world may end tonight? True, but on available evidence most of us make ready to commute on the 8:30 next day47 (p. 300). 160 ANNALS NEW YORK ACADEMY OF SCIENCES REFERENCES 1. Office of the Inspector General US N uclear Regulatory Commission. NRC's Regulation of Davis-Besse Regarding Damage to the Reactor Vessel Head. Case No. 02-03S, Dec 30, 2002. 2. Michaels, D. 2005. Doubt is their product. Sci. Am. 292: 96-101. 3. Smoking and health proposal. Brown & Williamson Document No. 332506. Avail able at http://legacy.library.ucsf.edu/tid/rgy93foo, accessed July 5, 2006. 4. Michaels, D. & C. Monforton. 2005. Manufacturing uncertainty: contested sci ence and the protection of the public's health and environment. Am. J Publ Health 95: S39-S48. 5. Brennan, D. Tobacco Institute, in'a 1989 interview on ABC television's Good Morning America. 6. Thompson, C. Memorandum to William Kloepfer, Jr. and the Tobacco Institute, Inc. from Hill andKnowlton, Inc.; October 18,1968. Tobacco Institute Document TIMN0071488-1491. 7. National Academy of Sciences, 2003. Planning climate and global change re search: a review of the draft U.S. climate change science program strategic plan. National Academies Press. Washington DC. 8. Luntz, E Memo on the environment. Available at Environmental Working Group: http://www.ewg.org: 16080/briefings/luntzmemo/pdf/LuntzResearch_ environment.pdf, accessed July 5, 2006. 9. N ational Toxicology Program, Board of Scientific Counselors. 2000. Summary of minutes from the report on Carcinogens Subcommittee Meet ing, December 13-15. Available at: http://ntp-server.niehs.nih.gov/ntp/htdocs/ Liason/121300.pdf, accessed July 5, 2006. 10. Eisenbud, M. 1990. An Environmental Odyssey: people, Pollution, and Politics in the Life of a Practical Scientist. University of Washington Press. Seattle, WA. 11. Sterner, J.H. & M. Eisenbud. 1951. Epidemiology of beryllium intoxication. Arch. Industr. Hyg. Occup. Med. 4: 123-151. 12. N ational Institute for Occupational Safety and H ealth, Criteria for a recom mended standard to beryllium exposure. DHEW HSM 72-10268 1972: IV-21. 13. January 1997 Beryllium Public Forum Albuquerque, NM. See www.eh.doe.gov/ be/forumal.htm 14. Efficacy of the 2 ng/m3 Standard. Exhibit B, Document ID CB053353. December 1991. 15. Paustenbach, D.J., A.K. Madl & J.F. Greene. 2001. Identifying an appropriate occupational exposure limit (OEL) for beryllium: data gaps and current research initiatives. Appl. Occup. Environ. Hyg. 16: 527-538. 16. Jeffress, C.N. Letter to Peter Brush, Acting Assistant Secretary, DOE. August 27, 1998. 17. U. S. Department of Labor. 1998. Unified Regulatory Agenda. 63 Federal Reg ister 22218, April 21. 18. Occupational Safety and H ealth Administration. 2002. Occupational ex posure to beryllium: request for information. 67 Federal Register 228: 70707, November 26. 19. Stange, A.W., D.E. H ilmas, F.J. Furman, etal. 2001. Beryllium sensitization and chronic beryllium disease at a former nuclear weapons facility. Appl. Occup. Environ. Hyg. 16: 405-417. MICHAELS: MANUFACTURED UNCERTAINTY 161 20. Henneberger, P.K., D. Cumro, D.D. Deubner, et at. 2001. Beryllium sensitization and disease among long-term and short-term workers in a beryllium ceramics plant. Int. Arch. Occup. Environ. Health 74: 167-176. 21. Kelleher, P.C., J.W. Martyny, M.M. Mroz, et at. 2001. Beryllium particulate exposure and disease relations in a beryllium machining plant. J. Occ. Environ. Med. 43: 238-249. 22. Rosenman, K., V Hertzberg, C. Rice, etal. 2005. Chronic beryllium disease and sensitization at a beryllium processing facility. Environ. Health Perspect. 113: 1366-1372. 23. International A gency for Research on Cancer. 1993. Beryllium, cadmium, mercury, and exposures in the glass manufacturing industry. Beryll. Beryll. Comp. 58:41-117. ' 24. N ational Toxicology Program. 2002.10th Report on Carcinogens. Available at: http://ntp.niehs.nih.gov/ntp/roc/eleventh/profiles/s022bery.pdf, accessed July 5, 2006. .25. Steenland, K. & E. Ward. 1991. Lung cancer incidence among patients with beryllium disease: a cohort mortality study. J. Natl. Cancer Inst. 83: 1380-- 1385. 26. Ward, E., A. Okun, A. Ruder, et at. 1992. A mortality study of workers at seven beryllium processing plants. Am. J. Ind. Med. 22: 885-904. 27. Sanderson, W.T., E.M. Ward, K. Steenland, et at. 2001. Lung cancer case- control study of beryllium workers. Am. J. Ind. Med. 39: 133-144. 28. Levy, P.S., H.D. Roth, et at. 2002. Beryllium and lung cancer: a reanalysis of a NIOSH cohort mortality study. Inhal. Toxicol. 14: 1003-1015. 29. Sack, K. & A. Mundy. A dose ofdenial. Los Angeles Times. March 28,2004. Avail able at: http://www.latimes.com/news/nationworld/nation/la-na-ppa28mar28- 1,1,4339482,print.htmlstory?coll=la-home-headlines&ctrack= 1&cset=true. Accessed October 10, 2005. . 30. Kernan, W.N., C.M. Viscoli, et at. 2000. Phenylpropanolamine and the risk of hemorrhagic stroke. N. Engl. J. Med. 343: 1826-1832. 31. LaGrenade, L., P. Nourjah, et at. 2001. Estimating public health impact of ad verse drug events in pharmacoepidemiology: phenylpropanolamine and hem orrhagic stroke. Poster presentation at the 2001 FDA Science Forum: Science across the boundaries. Washington DC, February 15-16. 32. Kirton, W. Email to Bayer representatives, SUBJECT: CHPA Yale Study Meet ing, 1/21/99. Available through the Los Angeles Times, "A Dose of De nial" at http://www.latimes.com/news/nationworld/nation/la-na-ppa28mar28- 1,1,2552623.htmlstory?coll=la-home-headlines. Accessibility verified October 10,2005. 33. Weinberg Group. Adverse event linked to OTC product, was available at: http://www.weinberggroup.com. Accessibility verified: July 30, 2004. 34. Krimsky, S. 2003. Science in the private interest: has the lure of profits corrupted the virtue of biomedical research? Rowman-Littlefield Publishing. Lanham, MD. ' 35. K rimsky, S. 2005. The funding effect in science and its implications for the judi ciary. J. Law Pol. 8: 43-68. 36. Sm ith, R. 2005. Medical journals are an extension of the marketing arm of phar maceutical companies. PLoS Med. 2: el38. 37. Bekelman, J.E., Y. Li & C.P. Gross. 2003. Scope and impact of financial conflicts of interest in biomedical research: asystematic review. JAMA 289: 454-465. 162 ANNALS NEW YORK ACADEMY OF SCIENCES 38. Lexchin, X, L.A. Bero, et al. 2003. Pharmaceutical industry sponsorship and ' research outcome and quality. BMJ 326: 1167-1170. 39. Gray, G.M., XT. Cohen, et al. 2004. Weight of the evidence evaluation of low- dose reproductive and developmental effects of bisphenol A. Hum. Ecol. Risk Assess. 10: 875-921. 40. Vom Saal, F.S. & C. Hughes. 2005. An extensive new literature concerning low- dose effects of bisphenol A shows the need for a new risk assessment. Environ. Health Perspect. 113: 926-933. 41. Bombardier, C., L. Laine, et a l 2000. Comparison of upper gastrointestinal toxicity and rofecoxib and naproxen in patients with rheumatoid arthritis. N. Engl. X Med. 343: 1520-1528. 42. Graham, D. Testimony before the U.S. Senate Finance Committee, November 18, 2004. Available at: http://finance.senate.gov/hearings/testimony/2004test/ 111804dgtest.pdf. Accessibility verified: October 10, 2005. 43. Berenson, A. 2005. For Merck, Vioxx paper trail won't go away. New York Times. August 21. 44. Davidoff, E, C.D. DeAngelis, XF. Drazen, etal. 2001. Sponsorship, authorship, and accountability. JAMA 286: 1232-1234. 45. Michaels, D. & W. Wagner. 2003. Disclosures in regulatory science. Science 3 0 2 :2073. . 46. Wagner, W. & D.M. Michaels. 2004. Equal treatment for regulatory science: extending the controls governing public research to private research. J. Law Med. 30: 119-154. 47. H ill, A.B. 1965. The environment and disease: association or causation? Proc. Royal Soc. Med. 58: 295-300.