INDUSTRY NEWS

SIA backs fab cancer study critics have long demanded

For several years some public health experts have cited anecdotal evidence that the chemicals used in chipmaking may cause cancer in fab employees, and they've pressed the semiconductor industry to conduct a cancer risk study. Chipmakers have pressed back, asserting that their employees work in one of the safest industries and that disparate recordkeeping practices over several decades make it difficult to reach a valid conclusion based on isolated cases.

In March the trade organization that represents almost 85% of U.S. semiconductor production decided it would try to put the matter to rest. The Semiconductor Industry Association (SIA) announced that it is authorizing a retrospective epidemiological study to determine whether U.S. fab employees have higher cancer rates than workers in other industries.

RISK FACTOR: SIA has authorized a retrospective study to see whether U.S. fab workers have higher cancer rates than workers in other industries. PHOTO COURTESY OF ATMI

"This industry has always looked for ways to improve the manufacturing processes that lower environmental impacts and improve the health and safety conditions for our employees," said George Scalise, SIA's president, in a news release carrying the announcement.

The key to SIA's decision is a finding by Johns Hopkins University that such a study is scientifically feasible. The trade association had commissioned the university's Bloomberg School of Public Health to sift through documentation from member firms. Following an 11-month investigation, the university team said member companies had provided enough historical data from their workplaces to show that a solid study was indeed possible.

An outside scientific advisory committee established by SIA in 1999 had recommended that the association conduct the preliminary review to determine the feasibility of a cancer study. The committee determined that there was "no affirmative evidence of increased cancer risk among U.S. semiconductor factory workers," SIA noted when it published the committee's recommendations in March 2002.

The scientific committee was led by David Wegman, chairman of the Department of Work Environment at the University of Massachusetts, Lowell, College of Engineering. The committee comprised eight experts in the fields of medicine, epidemiology, and toxicology, SIA says. Two representatives from the National Institute for Occupational Safety and Health also sat on the panel.

According to SIA, Genevieve Matanoski, the principal investigator for the Johns Hopkins team, said that after a "thorough examination of records from SIA member companies. . . we concluded that there is sufficient, reliable, and relevant data to conduct a scientifically valid retrospective study of cancer risks in the semiconductor industry."

Will the decision mollify critics who have accused the industry of stalling or engaging in public relations spin? A long-standing skeptic of the industry's public health record had only faint praise.

"Well, there's something positive to be said for it, and there's something negative," says Joseph LaDou, MD, director of the International Center for Occupational Medicine at the University of California, San Francisco (UCSF). "The positive thing is that someone somewhere is moving forward with a study. The negative is that SIA already has demonstrated in the past its inclination to control its research. They did it with their reproductive study. There's every reason to believe they'll do it with this study."

A UC Davis study sponsored by the association in 1989 resulted in the phasing out of ethylene-based glycol ethers because of their negative effect on women's reproductive health, the association says. However, the Silicon Valley Toxics Coalition (SVTC), a nonprofit watchdog and environmental advocacy organization based in San Jose, claims SIA's study was not comprehensive, examining only miscarriages and ignoring cancer rates and birth defects.

The coalition has consistently lobbied SIA to conduct a cancer risk study. In February 2002 the SVTC sent a letter to SIA and Wegman formally requesting information developed by the scientific committee. In its letter the coalition said the association missed a chance to determine the nature and extent of cancer and birth defect risks facing cleanroom employees, who work with toxic chemicals such as arsine, benzene, and hydrochloric acid.

In addition to conducting the preliminary review for an epidemiological study, SIA has said it will implement several of the scientific committee's recommendations. These include developing common job descriptions and language for maintaining relevant data, instituting a screening process for new semiconductor materials, and requiring chemical suppliers to provide toxicology tests. All these steps surpass the committee's purview, the association says.

Matanoski did not return phone calls from MICRO seeking comment. John Greenagle, SIA's communications director, asserts that the industry's critics have been unduly harsh.

Asked to clarify what Matanoski means by "sufficient historical data," Greenagle explains how difficult it is to collect and collate workplace information that is several decades old. "One of the challenges of doing a retrospective study, rather than tracking information going forward and then looking back, is to try to get at some of these records, which, by our industry's standards, are ancient history."

The team of experts that SIA eventually hires to conduct the study will examine "employee health records and exposure records, if there are any chemicals in use and the processes in use, the equipment" and related areas, Greenagle says. The formats containing the information will also differ. In its earliest forms some of the records will be in hard-copy form, others will be stored on 5 1/4-in. floppy disks. In addition, in multicompany studies "different job descriptions are in use at different companies."

"The biggest challenge to going ahead with the study was to determine whether you could collect the kind of records that would make it possible to do a study," Greenagle points out. "That's what Johns Hopkins spent 11 months doing. They had a number of site visits, met officials from a number of companies, and tried to look at the [records'] availability."

Given these difficulties, does it make sense to even try to conduct a study? Why not in good faith track the health and safety records of employees currently using chemicals in the industry's cleanrooms? Greenagle says the association had several reasons.

He notes, for instance, that the association plans to conduct the study "in spite of the fact that a prior panel, the scientific advisory committee, had looked at all of the available data and said, 'We find no evidence there that there's a higher risk or higher rate of cancer among cleanroom workers than among workers in the general population.' " The SIA critics, he insists, are trying to force its members into the classic conundrum of trying to "prove a negative." Nevertheless, "there were enough questions raised to do a further study, provided that the data would be available to draw a conclusion."

Greenagle joined SIA as communications director earlier this year. As a public relations specialist for 18 years with AMD starting in 1984, he recalls cleanroom tours where viewers would look through windows and "see wet sinks with chemicals being used. Today, you don't see any of that." Virtually all chemical processes take place inside sealed equipment, he points out. Therefore, "if there were a risk, it would have been higher in the earlier days when the practices hadn't evolved."

Another factor in the decision to proceed is that cancer takes a long time to develop. A final reason goes to the nature of the industry itself, insists Greenagle, who notes that he is a charter member of SIA's communications committee. A strong consensus had developed among members that "an essential part of this industry has been about finding what the facts are and if there is a problem, the industry wants to know about it."

Despite many examples of environmental, safety, and health programs and practices that the industry would argue are good-faith efforts, LaDou and other critics insist there needs to be more transparency in conducting studies to show that the industry is not trying to control the outcomes. As an example, the public health expert, who has provided data based on his work at UCSF to the toxics coalition, claims IBM has succeeded in preventing the publication of a valid scientific paper based on the chipmaker's corporate mortality file and analyzed by epidemiologists and Boston University. "They then proceeded with their own group of scientists at the University of Alabama, Birmingham, to do a study, the results of which we may not see for many years, if ever, while at the same time there's a pressing study done in the past that has embarrassing results that the company doesn't want to deal with."

LaDou and others critical of the industry's record are particularly annoyed by SIA's insistence on citing data from the U.S. Department of Labor, Bureau of Labor Statistics (BLS), concerning worker safety. The trade association has consistently boasted of its low injury rate as shown in BLS reports. In a recent news release, SIA cited a 2002 bureau report that ranked semiconductor manufacturers better than 95% "of all other durable goods manufacturing industries surveyed." The incidence of work injuries and illnesses was "only 1.9 per 100 full-time workers," the association noted.

LaDou points out that the BLS data were not unexpected at the time because "occupational injuries in all industries are six or seven times more common than occupational illnesses. An industry can double its rate of illness, but the increase will go unnoticed if the sum of the injuries and illnesses remains low."

He insists that the reason chipmakers have a low workplace injury rate is that less than 25% of their employees are production workers and the industry is a light-manufacturing business. Semiconductor manufacturing has twice the incidence rate of occupational illnesses "of any other manufacturing industry, even chemical and pesticide manufacturing. The rate of occupational illness related to exposure to hazardous and allergenic materials is even higher."

Greenagle concedes the point that the BLS data "show no relevance as to whether there's a cancer risk," even as he insists that BLS definitions of workplace illness and injuries often defy common sense anyway. "So, the study that will be done now should at least provide a basis for further discussion. Some of our critics are coming up with all kinds of lurid stories with no proof."

As of mid-April SIA had yet to receive a final report from Matanoski's team at Johns Hopkins, which is consulting with SIA's project management committee's internal group and the science advisory board, Greenagle says. The association is putting together a proposal request for contractors so that SIA can select a team by the end of 2004. Johns Hopkins says the report could take three to five years, Greenagle notes. "In my opinion we could have done more to try to explain to people the challenge of doing the study."

Does LaDou think it fair to admit there may be some difficulty in gathering solid worker health information going back decades from so many companies? "Could be," he responds after a moment's pause. "That certainly didn't prevent the companies from participating in the SIA study on miscarriages," pointing out that the industry had no hard information that glycol ethers necessarily presented a real reproductive threat. "There was some suggestion that removing glycol ethers from the workplace would be a good thing to do, but there was no proof that glycol ethers were the explanation for the problem, and so we suggested the SIA repeat this study.

"It could be done very quickly—to repeat a small reproductive study to demonstrate that removing glycol ethers reversed this miscarriage problem—and here it is seven to eight years later and they're still boasting that they did something way back when, but we don't have any idea what impact it's had on the public health. That's what's important."

As for the cancer study, LaDou says that a timetable of three to five years means "they must be planning a fairly extensive study. It's in their best interest to take as much time as possible. I would be disappointed if it's a small study. It would be awfully nice," he emphasizes finally, "if the scientific community could study the proposal." —JC