Concern over occupational health problems and environmental risks and increasing interest in marketplace health promotion programs, management science, work psychology, and other related subjects have led to an increasing number of proposals to conduct studies within a plant or in a specific workplace. Often the initial reaction of the plant manager and the responsible corporate officer is simple and quick: "no way!". Such a snap reaction often misses the key end benefits of allowing a study to be performed, however. To make a reasonable decision in the interests of the employer and all concerned, the corporate officer and the occupational health consultant must know the following:

1. Is the study proposed a worthwhile research effort?
2. Are the investigators reputable and capable of conducting the research objectively?
3. How can the company ethically monitor the course of the study to ensure that it remains within agreed-upon limits?
4. What are the potential benefits of participation to the company?
5. What will happen if a serious problem is uncovered?

The procedure outlined here is a guide to evaluating a serious research proposal. The emphasis is on studies in the field of occupational health but the same principles hold for studies involving environmental hazards, work psychology, and management science.

The first step in evaluating a proposed study is to determine whether a specific question is being asked. There are two general types of research, hypothesis-generating studies and hypothesis-testing studies. Hypotheses are statements of supposition, such as "Foundry workers have a higher rate of cancer than would be expected from rates in the general population". They are meant to be tested against actual data pertinent to the problem, such as cancer rates for foundry workers from a given plant compared to local area rates. A hypothesis-generating study is a fishing expedition, a superficial overview intended to spot possible leads worthy of further research. A hypothesis-testing study is a targeted attempt to confirm or rule out a specific hypothesis. A hypothesis-testing study is usually based on a previous observation, a logical extension of other research, or a previous hypothesis-generating study.

Without question, hypothesis-generating studies cause the most trouble. This is true because the results are likely on the basis of chance alone to suggest some problem even when there is none, since frequently dozens of comparisons are made and the traditional scientific standard for assuming significance allows a one-in-twenty chance of error. Furthermore, there is much confusion in the minds of the public and even many scientists regarding the interpretation of positive findings. Unless an association is very strong and the study is extremely well-designed, hypothesis-generating studies only suggest a possible finding; they rarely prove anything. To be sure of the finding, a hypothesis-testing study usually must be designed to look specifically at the problem of interest. Thus, hypothesis-generating studies are often crude, preliminary, and potentially misleading while hypothesis-testing studies are usually more refined, authoritative, and carry more credibility.

The second step in evaluating a study is to determine whether a topic is worth studying. This is often not as easy as it sounds because sometimes a trivial-sounding project is actually a test of a much more important general theory. When in doubt, it may be worthwhile obtaining an outside opinion from a trade association or a university. Here an astute manager may also see an opportunity to gain useful information. For example, if an investigator wishes to look at the health experience of employees in a specific work area of a self-insured plant, the employer may persuade the investigator to gather data that will help in containing health care costs and in evaluating the insurance plan.

The third step in evaluating a study is to determine whether the question being asked can actually be answered by the methods the investigator chooses to employ. Research design is a very complex scientific issue but at the very least a scientist proposing a serious research project should be able to explain what techniques will be used and how accurate they are, how many subjects will be needed to yield an interpretable result either way, and what statistical tests will be used to analyze the findings. Again, outside help may be needed.

Scientists have an ethical responsibility to maintain objectivity when they conduct and report their research. When a scientist consistently publishes conclusions for or against industry, it may mean that the investigator is biased or that the line of research actually led in that direction. The best guide to objectivity is what a scientist has published: Are all possible explanations for a finding spelled out? Is industry consistently painted as the "good guy" or the "bad guy"? Is the tone of the writing thoughtful or provocative? Can the scientist separate his or her political or social feelings from the technical and scientific aspects of a problem? The most highly respected scientists are usually those who are least predictable on the basis of partisan association, who make up their minds on the merits of each issue. Neither a study broadly inpugning industry by a radical activist nor a study broadly asserting that no problems exist by a corporate consultant are likely to be viewed as credible within the scientific community.

The credentials of the scientist are very important. Research is a serious and difficult business and those who undertake to do it must be well trained and competent; that usually means a doctoral-level degree from a recognized university. In the case of graduate student performing research for a dissertation, the credentials of the student's faculty advisor should show evidence of substantial academic accomplishment. Unlike business, there is no "school of hard knocks" in research; paper credentials should be complete (with no suspicious unaccounted-for gaps), commensurate with the scientist's age and experience, and show professional growth. A scientist who has not been promoted from a junior rank for many years, who has published very little recently, or who has moved frequently from one institution to another may be a bad risk. Such individuals may not be careful researchers or may feel under great pressure to blow a problem out of proportion in order to advance their stalled careers.

Where a scientist now works is an important factor. Universities have the highest credibility, in part because an investigator is under scrutiny by peers and students. Private research foundations vary from the impeccable to the fly-by-night and should have an airtight justification for why they are involved in the work. Government agencies often have trouble attracting and holding the best researchers. In every case, an institutional guarantee should be obtained to assure that confidential information will be kept under wraps until released in harmless form, trade secrets will be respected, access to the plant will be limited to authorized personnel, and all findings will be divulged to the company before publication. Penalties for violating safeguards should be spelled out.

If an investigator claims to have an affiliation with a university but cannot produce a university office address and telephone number, call the appropriate university department and find out the specific nature of his or her appointment. Adjunct, however, or part-time faculty often become involved in legitimate research projects but when they do, one wants to know where they work full-time and whether there is a potential conflict of interest with any other responsibilities they may have. Sometimes individuals with only tenuous university links to the institution are automatically reappointed to the adjunct faculty year after year, but do not have any substantial role on the faculty. Such persons have been known to use their courtesy appointments to lend authority to strictly personal projects. Beware of uninformative or unusual titles, such as "lecturer" (usually the lowest faculty rank and a faculty member not eligible for tenure) or strictly laboratory or administrative titles such as "research associate" or "director", unless the person can also show a proper faculty appointment or sponsorship of the project. The designation "in residence" usually means that a scientist supports him- or herself at the university by obtaining grants which cover salary and should not in itself be cause for concern. A legitimate investigator will always welcome inquiries into his or her background. A questionable risk may become defensive or may present an incomplete or confusing resume or "curriculum vitae", as it is called in academic circles.

Another important factor to consider is where the investigator is obtaining funding the or the research. If the funds have already been awarded by a research agency such as the National Institute of Occupational Safety and Health or if the investigator plans to submit a proposal to such an agency, some assurance is built-in that the proposal will receive careful review by other qualified scientists. Application to a recognized agency carries some assurance that the proposal will be reviewed for safety and ethical treatment of human subjects, if direct contact with workers is involved. If the funding comes from a private foundation or other unusual source, what is their ultimate purpose? When funds come from a trade association or industry source, is the study likely to be accused of pro-industry bias when it is released?

Finally, the attitude of the investigator toward publication is important. Responsible investigators will permit the company to review data before release for publication but will always refuse on ethical grounds to allow the company to edit the resulting paper or to block publication. Questionable applicants may compromise on these issues.

A company that allows research to be performed in its facilities has certain ethical and legal rights and obligations. The obligations are largely written into law but the rights are more vague except in the matter of protecting trade secrets.

The participating employer has the right to stipulate that trade secrets will be kept inviolable and can seek legal redress if an investigator or a research technician steps into expressly forbidden territory. The employer is entitled as an ethical matter to be briefed on the findings of a study, including any and all problems uncovered, before a study is released for publication. It is usually best to specify a time period, such as three to six months, between the briefing and release for publication. This allows time for management to take action to resolve any major problems and to minimize any public relations impact.

The employer is entitled to minimize the disruption in routine that the study may cause. The study protocol (procedures to be followed) should be kept as short as possible and should involve only as many employees as are necessary to answer the question.

The corporate culture is very different from that of academia. The corporate officer must make sure that plant managers and supervisors know the legal constraints on the employer and the ethical bounds of the researcher. Serious conflicts emerge when uninformed supervisors demand (in violation of the law) to see data that an investigator has collected before such data are statistically analyzed and all personal identification is removed. Even the rumor that confidential data might be released to an employer is enough to ruin the study and destroy any possibility of cooperation by employee's. Confidentiality is by far the most sensitive issue in occupational research.

The employer also has a right to insist on anonymity. In many cases the management will want to publicize the results because they may be favorable or, if neutral, may project an image of openness and corporate responsibility. Nonetheless, if the employer wishes to remain unidentified in the published report, it has that right.

The issue of employee participation is often troublesome for management but practical guidelines have emerged from long experience. Some studies require direct participation by workers. These studies are better received and are less likely to provoke conflict when employees are informed about the purpose, methods, and safeguards to confidentiality of a study at the very beginning. This usually means making a presentation to the union or employee representatives and obtaining their approval after management has accepted the research proposal. Employees have an absolute right to be given their own results and an explanation of the study's findings in terms a layman can understand. In cases where a serious problem has been uncovered, this step alone has proven invaluable in preserving labor-management relations, controlling rumors, and protecting the company from charges of cover-up. All such studies must be reviewed and approved by committees charged with protecting the safety and rights of human subjects in research.

With all the potential for problems, why should the corporate manager ever permit research to be conducted at all? The answer lies in the enormous benefit the employer can gain.

Most studies, of course, uncover no dramatic problems and may show exemplary conditions. Such outcomes can reassure worried employees, project an image of corporate responsibility, and provide evidence for compliance with governmental regulations.

Studies often generate information of great usefulness for management purposes, even when the data are collected for other purposes. Health data can be used to identify causes of absenteeism and controllable health care costs. Program evaluations can improve efficiency and performance.

In many cases, participation in a study may improve employee relationships. Where a demonstration project such as a health promotion program is highly visible, it may boost morale and a sense of identification with the company. There is also the familiar "Hawthorne effect", in which the mere fact of being studied boosted employee productivity by creating a sense of group identification and special attention among the employees serving as subjects.

Participation in such studies provides an opportunity to cement relationships with the academic community. In an era when joint ventures between academia and industry are increasingly common and are yielding profitable results, that is no small benefit. While academics are frequently accused of being too theoretical and impractical in their thinking, they cannot be otherwise if they are not allowed to participate in solving problems in the real world. A university connection provides opportunities for new product and service ideas, continuing education for technical and management personnel, and consultants at a reasonable cost.

The identification of a serious health problem is the nightmare of management in permitting research. Yet should this happen, the company that uncovers the problem first has the advantage of time to correct the problem and to limit the public relations damage, a strong legal position that due caution was exercised, and a clear defence against charges of coverup. An employer that would rather not know risks charges of neglect and duplicity at best and screaming headlines and a cascade of litigation at worst. This is especially true if there has been some rumor or intimation of a problem prior to the proposed study. Workers' compensation claims may be increased if a problem is uncovered, but if no data are available to determine the actual magnitude of the risk, which workers are affected, and what exposures may have been responsible, a smart lawyer will parlay the resulting uncertainty into a bevy of awards for all comers.

The key to managing a company's participation in a study is to understand the research process. The initial proposal should be studied carefully, with the help of outside consultants if necessary. A formal presentation by the investigator to management is an absolute essential. A decision on whether to participate should be made after the study is thoroughly digested and its potential benefit to the employer has been thought out. When a company receives a number of such proposals, it is best to hire a small standing committee to screen the studies and to make appropriate recommendations. The worst way to handle a proposal is to give it five minutes in the crowded agenda of a board meeting.

A written agreement between the investigator and the company is essential. This agreement should spell out all expectations and responsibilities on both sides, the measures to be taken to safeguard trade secrets and the confidentiality of employee data, penalties for violating the ground rules (particularly important if graduate students are involved), and guidelines for contact with employees. The agreement cannot require the investigator to withhold, censor, or edit the report of the research, but should explicitly give the company the opportunity to review the findings before they are released and to respond to any problems identified or implied criticism.

While the study is being implemented, management has a right to be kept informed on its progress but cannot obtain the data being collected. Once committed to participate in a study, a company must ride it out to the end. To unilaterally cancel a study midway is a disaster for all concerned -- for the employer it means a biased outcome, charges of coverup, and a public relations fiasco. The best way to manage a company's participation is to assign responsibility for monitoring its progress to one corporate officer who informs him- or herself about it in some depth and will thus be prepared to respond to the results when they are presented.

Sometimes an employer inadvertently kills a study with kindness. To remain credible to employees and as a scientist, the investigator must remain an outsider. This may mean declining invitations to eat in the executive dining room or offers of clerical assistance. Depending on the situation, management should not feel rebuffed if a scientist prefers to keep some distance.

Frequently an employer will ask for assistance to resolve a problem or to classify a source of potential liability. In such cases consultants are brought in under contract and the manager has the right to impose more rigid conditions, short of concealing a health hazard. Whatever the company can do in principle, however, it is most wise to treat such situations by the same standards as those outlined here for externally-initiated research.

Industry-based research, other than product development, has a tarnished reputation because it is susceptible or appears to be susceptible to self-serving pressure and manipulation. Academic standards for research, whatever its inconveniences and risks, have evolved to protect the credibility of the investigator from charges of conflict of interest and coercion. Where industry-based research is most prone to embarrassment, it stands to gain the most by opening the process. If it is not conducted and disseminated in keeping with academic standards, industry-based research will persist in lacking the persuasive ability to convincingly rebut contrary studies in the minds of the public and government regulators.

Cullen MR. Occupational Medicine. A new focus for general internal medicine. Arch Intern Med 1985; 145:511-515. 

Guidotti TL. Medicine, critical science, and the real world. West J Med 1975; 123:69-70. 

Guidotti TL. Exposure to hazard and individual risk: When occupational medicine gets personal. J Occup Med 1988; in press.