Tobacco Documents Online

3 Ctiu Epidemiol Vol. 44, Suppl. 1, pp. 29S-34S, 1991 0895-4356/91 53.00+0.00 Printed in Great Britain. AIl rights reserved Copyright C) 1991 Pergarnon Press p1c ETHICAL ISSUES RELATING TO THE CONDUCT AND INTERPRETATION OF EPIDEMIOLOGIC RESEARCH IN PRIVATE INDUSTRY GREGORY G. BOND Epidemiology, Health and Environmental Sciences, The Dow Chemical Company, 1803 Building, Midland, MI 48674, U.S.A. Abstract-This paper identifies the parties to which epidemiologists are obligated and explores the areas in which potential conflicts of obligations can occur. It is concluded that all epidemiologists, regardless of the organizational context in which they operate, share similar obligations to their subjects, society, their employer or funding agency, and their colleagues. The fundamental principals inherent to an epidemiologist's obligation to protect the public health is to identify real health problems and promote their resolution. It is in everyone's interest, including private industry, to respect those principals, and an industry supported epidemiology program can be an effective means for accomplishing that. These programs can be innovatively designed to protect those fundamental principles and thereby avoid conflicts of interest. This is illustrated with some examples from the epidemiology program operated at The Dow Chemical Company. Chemical industry Ethics Epidemiology Mortality Occupational illness Surveillance r- I it Soskolne [1] has recently suggested that a code of ethics be developed for epidemiologists. In doing so, he called attention to the potential for group interests to conflict with the pursuit of truth in epidemiologic practice. Such con- flicts have not been explored to any degree by epidemiologists, but need to be if we are to understand and avoid them. According to Beauchamp [2], a conflict of interest is a:,vays one in which a personal interest comes in conflict with a duty involving someone else's interest(s)--that is anyone else to whom you are obligated. It is closely associated with conflict of obligation, that is an obligation to two different parties who, in some fashion, are in disagreement. Thus, when assessing the potential for conflict of interest, two important questions need to be answered: (1) To which parties are epidemiologists obligated? and (2) Do the fundamental interests of those parties conflict, in particular do they conflict in the areas in which epidemiologists operate? This review will attempt to answer those two questions from the perspective of an epidemiol- ogist employed by a chemical company. In the process the following major messages will be developed: (1) All epidemiologists, regardless of their organizational context, have similar obligations to their subjects, to society, their employer or funding agency, and their colleagues. They also share similar personal obligations. (2) There are certain fundamental prin- (3) 29S ciples inherent to an epidemiologist's obligation to protect the public's health, viz. to identify real health problems, and promote their resolution. It is in industry's best interest to respect those principles, and an epidemiology

I 30S GnrGORY G. BoND program supported by industry can be an effective means for accomplishing that. (4) These programs can be innovatively designed to protect those fundamental principles and thereby avoid conflicts of interest. This will be illustrated with some examples from the epidemiology program operating at The Dow Chemical Company. Table 1 summarizes the obligations of epi- demiologists according to Beauchamp [2]. As health professionals, epidemiologists have certain obligations to their subjects. Paramount among them is to do no harm by word or deed, by commission or omission. But epidemiologists also have some obligations to communicate risk information in a timely and effective manner; to respect and protect the privacy of their study subjects; and to maintain the confidentiality of data entrusted to them. Potential conflicts can occur, for example, when in the course of trying to protect the privacy of their subjects and the confidentiality of data from those subjects, epidemiologists are asked by their employers to share data on an individual basis. Exactly the same conflicts can also occur when they are asked by other inves- tigators who are not a part of the study team to share this same data. Epidemiologists are obligated to society to avoid conflicts of interest; to avoid allowing their personal values to influence the results of their research, to share information-not necessarily data, but information, and to pursue their responsibilities with due diligence. They must also obey the spirit and letter of the law. As an aside, it should be noted that many epidemiologists who are employed in private industry are obligated under Section 8(e) of the Toxic Substances and Control Act [3] to communicate substantial risk information to the Administrator of the Environmental Protection Agency within 15 days of learning of that information. These notifications are then made available to the public. Failure to notify can involve substantial civil and criminal penalties. Although there is no disagreement with the intent of the law, the time demands can produce a potential conflict between one's legal and professional obligations. The professional obli- gations suggest that you must make your best interpretation of the facts, perhaps even to the extent of realizing that you cannot make the Table I Epidemiologists are obligated to: human subjects society funders and employers colleagues Adapted from Beauchamp [2]. best interpretation without additional facts. Professional obligations suggest you must not publically communicate a message that you know to be a half-truth. The legal obligation can say, no time for additional work; make your best interpretation now. Epidemiologists ate obligated to their funders, employers, and sponsors to specify their other obligations; to present all available alternatives and a framework for evaluating the merits of each; and to protect privileged information. A potential conflict exists here because some courts have maintained that the obligation to the employer supersedes certain obligations to the fellow employee-in particular, it super- sedes the obligation of confidentiality of infor- mation. There is currently a case ongoing in Pennsylvania, in which an occupational nurse refused to turn over to her employer certain employee laboratory results. She had been fired and was suing for wrongful discharge. The courts held that she was obligated to turn over the information. The American College of Occupational Medicine is helping her on the appeal. This particular conflict would have been avoided if her employer had been made aware of her other obligations and had respected those obligations. To their colleagues, epidemiologists are obligated to maintain public confidence in the discipline by promoting it while not overstating its capabilities; to adequately communicate methods and results; to expose unacceptable behavior or conditions whenever they are found; and to communicate ethical requirements. A potential conflict exists here if one epi- demiologist is asked to unduly criticize the work of another. The operative word is unduly. A blanket prohibition on criticism would be in conflict with the obligations to seek truth. Science proceeds by the stages of hypothesis, disagreement, and resolution. Anything that interferes with that process is to be avoided. And then there are the obligations that epi- demiologists necessarily feel to themselves and their families. Although they may be reluctant to acknowledge it, each feels obligated to I

Epidemiological Research in Private Industry provide their family with some level of financial support, and to remain faithful to their moral and religious teachings. Some might argue that any intrusion of personal values into the scien- tific process should be completely avoided. But this seems unrealistic. Instead, these personal values need to be recognized, acknowledged and used in a constructive fashion. Note that there are many organizations which have interests in epidemiology and public health to whom no special obligation is owed. For instance, epidemiologists have no duty, per se, to Ralph Nadar's Public Interest Research Group (PIRG). You can disagree with them and still be true to all of the other parties toward which you do have a duty. In fact, you may have an ethical imperative to disagree with PIRG if they have taken a position or done something that conflicts with your other obligations. For instance, last fall the Michigan chapter of PIRG (PIRGIM) held a news conference to release an ecological study they had sponsored by a group called Public Data Access [4]. The study alleged a relationship between toxic pollution in 13 Michigan counties and elevated cancer rates. Release of the study occurred 2 weeks prior to voting on a public referendum to increase fund- ing for environmental projects. Now Midland County, site of Dow's facilities, was not among the 13 cancer problem counties. However, I felt compelled, out of professional interest, to review the study anyway and found it to be greatly flawed. My critique was shared with others in the field including state public health officials. I wonder how many other epidemi- ologists also felt that PIRGIM had misused the epidemiology and did not speak up because they felt more comfortable in remaining anonymous. Did that not represent a conflict between their personal and professional obligations? What are the fundamental principles inherent to the obligations of epidemiologists? Gordis [5] described the objectives of epidemiologic research as enhancement of our understanding of the biology and pathogenesis of disease and the improvement of human health. He also emphasized the importance of identifying causal relations as opposed to merely identifying associations. Note that a fundamental principle is not the maintenance of a federal agencies' budget, helping a junior faculty member achieve tenure, providing a data set to a graduate student, or a whole variety of other things. These may be processes by which fundamental principles 31S are indirectly achieved, but they are not fundamental in and by themselves. This seems obvious, but some people frequently confuse the two. Organizations can also differ in their process and approach to issues even when they adhere to the same fundamental principles. As a conse- quence, one can disagree over process and still be true to the same fundamental principles. In such situations, there is no basic conflict. For example, we at Dow have had significant differ- ences of opinion with some of the federal health research agencies about process while still both adhering to the same fundamental principles. Some have wanted to tap into the Dow epidemi- ology databases, and when told that access was conditional upon receipt of an acceptable study protocol, a condition wc place on ourselves, they claimed we were being obstructionist and unethical. Fortunately, after some initial acrimony, cooler heads prevailed and a suitable resolution was achieved. As mentioned at the outset, I am an epidemi- ologist who works in industry. Many people y may feel that industry, including all who work there, have a different set of fundamental prin- ciples regarding the obligation to protect the public's health-i.e. protect the company at all costs by denying risk. I disagree. We feel that if we protect the public health, employee health, we are protecting the company. The two are not mutually exclusive. Paradoxically, we protect the company by aggressively trying to find real causal relationships, not by doing our best to deny their existence. The Epidemiology Department at Dow has a mission of providing timely information on the human health effects potentially associated with the agents used or produced by the company. People frequently ask me whether or not Dow management is truly interested in finding prob- lems. There is a saying in the company, "Don't look upon it as a problem. Look upon it as an opportunity." Many of the Dow managers are engineers or scientists. They were promoted to their positions because they have demonstrated an ability to solve problems in the face of adversity. They are not afraid of problems. What they uniformly do not like are surprises. So in designing an epidemiology program it was necessary to emphasize good communication. This will be elaborated on later. Presented first, however, is a description of the general philosophy behind our sequential approach to conducting research at Dow, which is based CEUUrttl-D

32S on maximizing both efficiency and scientific validity (see Fig. 1). Mortality, morbidity and medical surveil- lance databases have been assembled to allow a timely response to questions which are posed from outside the Dow epidemiology group. More importantly, they also allow Dow epi- demiologists to establish a proactive research agenda based on the health experience of the employees themselves. For instance, a census has been assembled of all present and former full-time employees of Dow Chemical U.S.A. since 1940. It is referred to as a Vital Status Registry and represents more than two million person-years at risk. The employee's vital status is con- tinuously being updated as follow-up is done using the records of the National Death Index, The Social Security Administration and others. Periodically, the mortality experience of all of the workers from a particular manufacturing site is examined in comparison to expected levels based on national, regional and local rates [6-9]. These general mortality surveys frequently include analyses by socioeconomic level, year and age at hire and duration of employment, but not by chemical exposure. It is not feasible at this stage, because of the large number of employees and the complexity of the personal and occupational exposure environment. Instead, causes of death which occur in excess are candi- dates for study with the case-control (or hybrid study) approach, wherein exposures need to be characterized only for the cases and for a sample of the remainder of the cohort [10, 11]. Because Work Histories and Exposure Profiles the focus is narrowed to one or several causes of death, it then becomes feasible to collect data on a particular set of potential confound- ing variables, thereby increasing the validity of the research. Note this company based sys- tem is geared to finding potential problems and to placing them in perspective. The key is the commitment to the whole cascading process. Cohorts of workers who have had the poten- tial for exposure to specific agents can be linked to these databases to accomplish agent-specific cohort studies in a timely manner [12, 13]. The timeliness aspect should be emphasized, because of its humaneness. Prior to ha;~ing assembled a surveillance system, it would take several years to respond to a question about the workers' health experience with a particular chemical- the studies would take that long to complete. In the meatime, there was considerable uncer- tainty and anxiety about the situation. Now, those questions can be answered in a matter of days or weeks rather than years, thus making epidemiology a more responsive tool to address concerns that may have arisen on the shop floor, in the boardroom or in the news media. In addition, the data is of better quality because certain validity checks can be done on the whole data set that would have been impossible on data subsets. Also note, having the data in hand permits not only reactive but also proactive research. If the program were not in place, the work simply would not have been possible. The mortality experience of Dow employees would be largely unknown. Vital Status Registry Division-Based Mortality Survey i Nested Case-Control Studies T Agent-Specific Cohort Studies Fig. 1. Sequential approach to mortality research at The Dow Chemical Company. ~ 1 iaeuef

Epiderniological Research in Private Industry A considerable amount of time was taken to understand the basic fundamental principles inherent to epidemiologists' obligation to pro- tect employee health before designing a pro- gram that embraced these principles. In the course of this evolution certain processes that were already operational in the company were utilized. In other cases, the processes had to be modified. For example, the research is no longer funded on a project by project basis. Instead, the whole program is funded. As a consequence, progress updates with preliminary results are not provided. Therefore, continuation of fund- ing is not contingent upon obtaining certain results. By modifying the process, a potential conflict of interest was designed out of the system. Collecang the data for a program rather than for individual study projects also ensures that it is done independent of any preconceived bias about a particular study hypothesis. On the other hand, having all of that employee data stored at the ready in the computer could possibly lead to it being used for inappropriate purposes. This is protected against by requiring an approved study protocol before access to the data can be obtained. Each study has a written protocol that details the authors, usually including at least two epidemiologists; an introduction and litera- ture review; a statement of the objectives; the methodology, including any survey instruments; a rationale for interpreting the results, including a discussion of how the study will deal with issues such as selection, misclassification, con- founding, and power; and costs of the study in terms of time, money and personnel. Each protocol is reviewed for technical merit by peers, and is reviewed with management so that they are aware, in advance, of all of the potential risks-remember, no surprises! Over 5 years ago the department formed an Institutional Review Board, which operates in full compliance with NIH guidelines, and includes members who represent the community. They perform the final review and approval of study protocols. They have the authority to deny approval to any project to to halt ongoing data collection if problems of an ethical nature arise. At the end of each project similar steps are gone through, and each study is completed. Every one has a report that is consistent with the protocol. The report is reviewed for technical merit by other scientists within the company, and then is presented to the management and 33S the employees. Furthermore, there is a commit- ment to publish or present at a scientific meet- ing, every study that is undertaken, regardless of the findings. Dow management is aware of and supports that commitment, and the depart- ment's record of more than 75 published articles during the past 10 years bears than out. To summarize, it must be emphasized that a company-based epidemiology program is an important aspect of an employer's obligations to its employees, customers, and society. Indeed, to consciously decide not to conduct such research might be interpreted as unethical. All epidemiologists, whether they are employed by academia, private consulting firms, hospitals, government agencies, public interest groups, or industry are under similar obligations and face potential conflicts of interest. This does not necessarily disqualify the individuals or their studies. The challenge to each of us as pro- fessionals is to anticipate potential conflicts of interest and then use innovative means for scrupously avoiding them. Based on my experi- ence, it is possible to develop programs that allow epidemiologists to fulfill their various obligations and do so with a minimum of conflict. REFERENCES 1. Soskolne CL. Epidemiology: Questions of science, ethics, morality and law. Am J Epidemio! 1989; 129: 1-18. 2. Beauchamp TL. Draft Code of Ethics for Epidemi- ologists. Kennedy Institute of Ethics, Georgetown University; 1989. 3. Toxic Substsnces Control Act-Law and Explanation. Chicago, Ill.: Commerce Clearing House, Inc.; 1976: 79-82. 4. Public Interest Research Group In Michigan (PIRGIM). The Cancer Connection: An Exploratory Analysis of the Relationship Between Toxic Waste and Cancer Mortality in Michigan. Ann Arbor, Mich.: PIRGIM; 1988. 5. Gordis I. Challenges to epidemiology ?n the next decade. Am J Epidemiol 1988; 128: 1-9. 6. Bond GG, Shellenberger RJ, Fishbeck WA, Cartmill JB, Lasich BJ, Wymer KT, Cook RR. Mortality among a large cohort of chemical manufacturing employees. J Natl Cancer Inst 1985; 75: 859-869. 7. Bond GG, McLaren EA, Cartmill JB, Wymer KT, Sobel W, Lipps TE, Cook RR. Cause-specific mor- tality among male chemical workers. Am J Industr Med 1987; 12: 353-383. 8. Bond GG, McLaren EA, Cartmill JB, Wymer KT, Lipps TE, Cook RR. Mortality among female em- ployees of a chemical company. Am J Industr Med 1987; 12: 563-578. 9. Olsen GW, Hearn S. Cook RR, Currier MF, Allen S. Mortality experience of a cohort of Louisiana chemical workers. J Occup Med 1989; 3 L• 32-34. 10. Bond GG, Shellenberger RJ, Flores GH, Cook RR, Fishbeck WA. A case-control study of renal cancer

34S GxEGORy G. BoND mortality at a Texas chemical plant. Am J Industr Med 1985; 7: 123-139. 11. Bond GG, Flores GH, Shellenberger RJ, Cartmill JB, 13. Fishbeck WA, Cook RR.Nested case-control study of lung cancer among chemical workers. Am J Epidemiol 1986; 124: 53-66. 12_ Hearn S, Ott MG, Kolesar RC, Cook RR. Mortality experience of employees with occupational exposure to DBCP. Arch Emiron Health 1984; 39: 49-55. Bond GG, Wetterstroem NH, Roush GJ, McLaren EA, Lipps TE, Cook RR. Cause-specific mortality among employees engaged in the manufacture, formu- lation or packaging of 2,4-dichlo rophenoxyace tic acid and related salts. Br J Industr Med 1988; 45: 98-105.