The vinyl chloride studies most often cited today — including a major study soon to be published — in fact are updates of a study first done in 1974. After companies learned of workers suffering from angiosarcoma, they quietly decided to find out what other cancers vinyl chloride might be causing.
The industry study was flawed from the start. The weaknesses built in to it only became worse as decisions were made on how to update it.
In June 1973, the industry’s trade group, then known as the Manufacturing Chemists’ Association, hired the consulting firm Tabershaw-Cooper Associates to tabulate cancers at vinyl-chloride plants. The first challenge was to compile a list of workers exposed. Rather than let scientists at Tabershaw-Cooper ultimately decide which workers should be put on the list, the chemical companies assigned the task to their own plant managers. At Union Carbide, managers decided to include only people working directly with vinyl chloride, based on some written records but also on supervisors’ distant memories.
Until the mid-1970s, exposure data was crude to non-existent. The managers reasoned that workers’ recollections of the potency of odors — categorized as high, medium or low — would be one way to estimate exposures. Jim Tarr, who worked as an air pollution regulator in Texas at the time, said such a method “doesn’t even reach the level of being junk science.”
Tarr, now an environmental consultant in Southern California, said it’s ridiculous to expect anyone to remember distinct odors years after the fact. In fact, vinyl chloride can be smelled only at levels far higher than even the old regulations allowed.
Tabershaw-Cooper’s final report — without revealing the methods used — said that measuring exposures at the plants “proved generally to be impossible.” It acknowledged that managers’ techniques for determining levels of exposure were “subjective” and had “questionable validity.”
Even with this problematic data, Tabershaw-Cooper reported in 1974 that there were more brain tumors than expected at vinyl chloride plants. A follow-up completed in 1978 reported that brain cancers at vinyl-chloride plants were occurring at twice the normal rate.
There was evidence from the start that Union Carbide workers in Texas City who died of brain cancer had been exposed to vinyl chloride. When news of the first 10 brain cancers at the plant broke in 1979, Union Carbide’s Gulf Coast medical director, Dr. David Glenn, acknowledged as much while also trying to deflect blame from the chemical.
"Although the press has strongly indicated that vinyl chloride may have been the culprit, only about one-half of our [brain cancer] cases had any known exposure to this chemical,” he said in a statement.
Yet none of those workers was included in the study updates that have formed the bedrock of today’s scientific consensus. The only brain cancer death from Texas City included in these updates was that of Luther Ott, a 57-year-old production worker who wasn’t even diagnosed until a month after the medical director’s statement. Ott died in February 1980.
Chemical industry officials knew before they hired Otto Wong to do an update that none of the 10 brain cancer deaths in Texas City had been included in previous studies, even though Glenn said half of the workers had been exposed to vinyl chloride.
One week after Glenn’s statement, Union Carbide’s corporate medical director, Dr. Mike Utidjian, told an industry task force that none of the 10 Texas City victims had a “clear cut” exposure. Nor were any included in previous studies.
Wong said it would have made more sense to start the study over rather than update a flawed one.
“From the scientific point of view, a better approach would be to do a new study,” he said.
That would entail reanalyzing which workers were exposed and which weren’t.
In fact, by March 1981, scientists at Union Carbide had determined that at least four of the workers who died of brain cancer had been exposed to vinyl chloride. The biostatistician who wrote that memo, Rob Schnatter, declined to comment for this story.
Schnatter did not keep the four dead workers a secret. He and another Union Carbide scientist acknowledged them in an article published in 1983.
Schnatter wanted to amend which workers were in the industry study. In 1982 he sent a memo to his colleagues at Union Carbide, one of whom wrote a handwritten response : “No, we are not adding people to the cohort.”
This reflected a critical decision that all but guaranteed the study’s outcome. According to the protocol, workers included in the original study could be dropped from updates if new information showed they hadn’t been exposed to vinyl chloride. But the reverse wasn’t true. Workers not initially included in the study couldn’t be added even if it turned out that they had been exposed, according to a Union Carbide memo.
In 1974, Tabershaw-Cooper was given a list of 431 exposed workers from Texas City. But when the study was updated a decade later, the number of exposed workers had dropped to 289 names.
Susan Austin, a Union Carbide epidemiologist at the time, complained in an internal memo that the odd rules for reclassifying whether workers were exposed “could lead to substantial bias.”
Collins, the former Dow epidemiologist, said it should been nearly impossible to cheat on this type of study. When scientists are deciding which workers were exposed to a chemical, they usually don’t know which ones have died. Therefore, they can’t skew the outcome by excluding dead workers.
“There’s no way to fudge the data,” Collins said.
But in this situation, Union Carbide did know which workers had died. It also knew it was excluding workers who had been exposed to vinyl chloride. The Center found no evidence that Union Carbide removed workers with brain cancer who had been in the original 1974 study. But the documents show that when the study was updated, at least three brain-cancer victims Union Carbide knew had been exposed were not included.
“It looks like they did leave them out by their own admission,” said former NIOSH official Lemen, who at one time served as a consultant for lawyer Baggett.
Kenneth Mundt, the lead author of the most recent update of the vinyl chloride study and a principal at the consulting firm Ramboll Environ, at first promised to answer questions from the Center. But weeks later, Mundt said that the study’s sponsor, the American Chemistry Council, wouldn’t allow him to talk because of pending litigation.
A Dow spokesperson said, “If Texas City workers met the eligibility criteria … then they would have been included in the industry-wide study, regardless of the cause of death …. Not all Texas City workers had opportunity for exposure to vinyl chloride.”