Breast Cancer Researchers Erred

A group of authors, including several Harvard faculty members, announced earlier this week that they erred in a study published this past July which concluded that the relationship between those who develop breast cancer and those who have a family history of the disease was weaker than previously believed.

In Wednesday's issue of the Journal of the American Medical Association (JAMA), the researchers corrected their original conclusion that only 2.5 percent of women with breast cancer have family histories of the disease. The conclusion should have read six percent, according to Dr. Walter C. Willett, a co-author of the study and chair of the Department of Nutrition at the School of Public Health.

Willett attributed the incorrect conclusion, published in the JAMA of July 21, to a mathematical error.

An eight member team, which included five researchers affiliated with the School of Public Health, conducted the original survey on nearly 120,000 women in the Nurses' Health Study.

Before it was published, the procedure and results of the research had gone through the journal's peer review process, in which experts advise thejournal whether a paper is worthy of publicationand what corrections, if any, should be made.

But the error was overlooked. "No one ever sayspeer review reflects pure accuracy," said Dr.George D. Lundberg, editor of JAMA.

"The error occurred in picking out data to plugin the formula," Willett said. Willett said theerror was only a minor part of the analysis. "Noneof the primary results was wrong."

Willett was critical of the Boston Globe'scoverage of the incident. He said they "made sucha thing out of it."

However, Lundberg, who is also an adjunctprofessor at the School of Public Health thisyear, said that it was a "significant error" andthat the mistake was "noted greatly in publicmedia because the result is different fromprevious ones."

But Karen M. Rouse, spokesperson for theMassachusetts chapter of the American CancerSociety, said there is "not such a hugediscrepancy" between the original result and therevised conclusion.

Willett said he felt that the correlationbetween breast cancer and family history of thedisease has been exaggerated. Although women withfamily history are twice as likely to develop thedisease, Rouse said that at least 90 percent ofthose women who develop the disease do not havefamily history of the disease.

Breast Surgeon Dr. Susan L. Troyan of FaulknerBreast Center agreed, saying that "most woman whoget breast cancer don't have a family history ofbreast cancer."

But the cancer specialists warned that thosewithout a family history of the disease shouldstill be alert to its dangers. "Any woman has thecapacity to develop breast cancer," Lundberg said.

Rouse suggested "some education for the publicon what research is about," implying that thepublic sometimes may not fully understand theresults of scientific studies.

Both Willett and Lundberg agreed that thepublic places too much expectation on scientificresearch. The two emphasized that science is an"ongoing process" and that no conclusion isabsolutely beyond doubt

But the error was overlooked. "No one ever sayspeer review reflects pure accuracy," said Dr.George D. Lundberg, editor of JAMA.

"The error occurred in picking out data to plugin the formula," Willett said. Willett said theerror was only a minor part of the analysis. "Noneof the primary results was wrong."

Willett was critical of the Boston Globe'scoverage of the incident. He said they "made sucha thing out of it."

However, Lundberg, who is also an adjunctprofessor at the School of Public Health thisyear, said that it was a "significant error" andthat the mistake was "noted greatly in publicmedia because the result is different fromprevious ones."

But Karen M. Rouse, spokesperson for theMassachusetts chapter of the American CancerSociety, said there is "not such a hugediscrepancy" between the original result and therevised conclusion.

Willett said he felt that the correlationbetween breast cancer and family history of thedisease has been exaggerated. Although women withfamily history are twice as likely to develop thedisease, Rouse said that at least 90 percent ofthose women who develop the disease do not havefamily history of the disease.

Breast Surgeon Dr. Susan L. Troyan of FaulknerBreast Center agreed, saying that "most woman whoget breast cancer don't have a family history ofbreast cancer."

But the cancer specialists warned that thosewithout a family history of the disease shouldstill be alert to its dangers. "Any woman has thecapacity to develop breast cancer," Lundberg said.

Rouse suggested "some education for the publicon what research is about," implying that thepublic sometimes may not fully understand theresults of scientific studies.

Both Willett and Lundberg agreed that thepublic places too much expectation on scientificresearch. The two emphasized that science is an"ongoing process" and that no conclusion isabsolutely beyond doubt