14 March 1999
PROJ99\POLMEDJ.MS
2,500 words

Teenage deaths linked to irresponsible press!
NEJM lead article faked, hacks fall for it
"How could you do this to us, Sally?"

Lundberg wasn't the only firing

NEW YORK--Enough Pulitzers! Who's winning the horse race for the Ignoble Prize in journalism? Medical reporters, of course--they can do more damage than anyone else.

When patients are faced with life-and-death decisions, journalists can and do give misinformation that drives readers in a panic to their deaths, writers heard at a Columbia J-school seminar on 13 March 1999.

Irresponsible stories in the U.K. encouraged patients to discontinue relatively safe contraceptive pills, for more dangerous alternatives like abortion, David Frankel, MD, explained at an all-day seminar, "Breakthrough? The Science and Politics of Medical Journalism." The problem was that they didn't understand the magnitude of the numbers and the limitation of the study.

Frankel, who is North American editor of The Lancet, emphasized the special responsibility of himself and all other medical journalists. "Patients take what we write and act on it," he said. "Medical journalists do in fact practice a form of medicine."

Journalists push cancer patients to demand drugs that may be ineffective or fatal, charged law professor Arthur Miller in his Socratic dialog. Gina Kolata was still being ridiculed.

Public health offers the greatest opportunities to do harm--to entire populations of the most helpless. Martin Schechter, MD, described how journalists with a political agenda ignored the facts and clear statements in the conclusion of his widely-misinterpreted paper on the Vancouver needle exchange program, in order to reverse his conclusions by 180 degrees. As a result, other researchers calculated that thousands of Americans have been needlessly infected with HIV, including perhaps half of all infants born with the disease.

Politicians said, "More people in needle exchange programs get HIV. Therefore, needle exchange programs cause HIV. Therefore let's shut needle exchange programs down." With equal logic, said Schechter, you could say, "People in hospitals are more likely to die. Therefore, let's shut hospitals down." The problem was that journalists--and right-wing politicians--didn't understand, and didn't try to understand, confounding variables.

NEJM, O&G, screw up royally

Fortunately, these problems mostly affect other people, not experienced medical reporters (like us), who rely on published, peer-reviewed studies in reliable journals like the New England Journal of Medicine. Which is why the audience woke up when Bruce B. Dan, MD, gave a few case histories from the NEJM. There was John Darsee, a young Harvard cardiologist with 46 publications (since retracted), who found a family with inherited cardiomyopathy. Dan showed the family tree on a page from the NEJM. A 17-year-old boy had an 8-year-old daughter. "Strange family," noted Dan. "Totally fabricated." Another distinguished journal, Obstetrics & Gynecology, found and characterized objects that they identified as a new organism in the blood of patients suffering from toxemia in pregnancy. The objects subsequently turned out to be lint on the cover slips in their laboratory. The NEJM also made mistakes that were a bit more subtle. Dan gave a review of case control studies, randomized controlled trials and other types of studies, with the strengths and weaknesses of each.

But even publishers who understand all the science can still do harm. Along with George Lundberg, Andrew Skolnick was also fired from JAMA. Skolnick, a reporter for "Medical News & Perspectives," won a $25,000 Goldsmith Prize for his coverage of incompetent prison doctors. The Alabama Medical Society felt that JAMA should stop reporting on social issues and concentrate on clinical medicine.

Fountain of truth

Ellen Ruppel Shell, best known for her New York Times Magazine article, "The Hippocratic Wars," warned that the medical journals are published by medical associations, earn substantial advertising revenues, have their own biases, they're published by falliable humans who are part of the club, and should be covered with the same skepticism that you cover anything else.

One of the inherent biases of medical journals is a bias towards publishing good results, said Shell. If 19 researchers submit papers with negative results, they won't get published, but if 1 researcher submits a paper with positive results, that will get published. Some tips for reporters: Check the foreign journals; they're at least part of a different club sometimes with a different perspective. Lundberg has become a revered, martyred figure, but Shell was skeptical. She told of Lundberg touting a forthcoming paper to her, saying, "Ellen, this is going to be hot." Shell didn't think the oral sex paper, for which Lundberg was fired, was important. She scoffed at publishing a paper on therapeutic touch by 11-year-old Emily. Skolnick dissented.

Law school seminar

The day ended with a Fred Friendly seminar, a Socratic debate named after the late Friendly, who resigned his post at CBS News after CBS chairman Bill Paley refused to interrupt the broadcast of "I Love Lucy" for the Fullbright hearings on whether the U.S. should get involved in the Vietnam war. After a furious argument, Friendly angrily quit, and stormed out the door--into Paley's private bathroom by mistake. Friendly then taught at Columbia. The seminars are moderated by Harvard law professor Arthur Miller, in the style of a law school case history, his outrage occasionally moderated as he slipped into a W.C. Fields voice.

In the hypothetical case history, a researcher at a big oncology conference, played by Schechter, discusses a study of a new drug for stage 4 breast cancer, now at 12 months of a 36-month trial, with 200 patients randomized into the treatment group, and 200 controls. After 18 months, the benefits are so dramatic (40 deaths vs. 80 deaths), despite some minor methodological flaws, that the institutional review board stops the trial. While the journal editors, played by Frankel and Dan, are arguing over whether and how to publish the paper, word slips out to newspapers and TV, which run the story of a dramatic new cancer breakthrough. Helen, a cancer patient with stage 3 breast cancer, hears about the drug on the media, and finds a doctor to give her the drug off-label, even though it was approved for a more advanced stage. 28 months out, 12 patients in the treatment group develop neutropenia and septic shock, compared to 3 in the control group. 52 compared to 83 is still pretty good, but not for Helen, who only had stage 3 anyway.

Oral contraceptives shocker

But back to the real case histories. Think of Gulliver's Travels. Imagine a trip to a parallel universe where people look like us, talk pretty much like us, and conduct scientific research just like we do--but all the newspapers are published by people like Rupert Murdoch.

This strange land is the United Kingdom. In 1995, 1.5 million women in the UK were taking oral contraceptives. "More is known about the safety of the pill than any other medication placed in the human mouth," said David A. Grimes, an authority quoted by Frankel. The most significant risk from oral contraceptives is deep vein thrombosis (DVT), the formation of blood clots in the legs, which sometimes leads to pulmonary embolism (PE), which is sometimes fatal. There are 2nd generation contraceptives and newer 3rd generation contraceptives.

According to the best data, the incidence of DVT is about 4/100,000 in the population of fertile women, 15-20/100,000 among women on the 2nd generation pills, 20-30/100,000 on the 3rd generation pills, and 60/100,000 among pregnant women. However, the incidence of death from PE is only about 1-2/100,000, compared to 200/100,000 for pregnancy. So the relative risk of DVT and death from the 3rd generation pills is about 2 times higher than the 2nd generation pills. But what kind of studies do these numbers come from?

In 1995, the World Health Organization did a trans-national study of women on the pill, and found that, in preliminary data, there was an increase in DVT. WHO forwarded the information to the British Government, which decided to forward the information to British physicians in what in the US would be called a "Dear Doctor" letter. The British government send the letter by mail, because they didn't consider it important enough to fax. The letter said that, according to 3 unpublished studies, the 3rd generation pill is associated with "around a 2-fold increase in risk." 3rd generation pills, the letter said, should be prescribed only "in patients intolerant of others," or those who "are prepard to accept the risk."

Dueling headlines

The press got the letter before the doctors did. Frankel told the story with a series of screaming headlines on the screen. "Teenage Deaths Linked to Pill," said the Sunday Times, in a story written by its legal correspondent. "Sixteen-year old Nancy Berry, a keen disco dancer, was fit and healthy--until she died."

"Official Birth Pill Warning," said the Evening Standard. "Pills that killed a young woman." The government tried to reassure the public: "Doctors Tell Women on 'Unsafe' Pill Not to Panic."

A professor H. Jacobs criticized the press coverage with the kind of colorful, if immoderate, quotes that American journalists can only dream of: "This is unutterable claptrap... it is nonsensical, meanminded speculation... from ignorant pigs."

"Try to picture women reading these newspapers," said Frankel.

Frankel told the story of Sally Hope, a doctor in a small rural community growing mostly "sheep and mud." Dr. Hope didn't know what was going on, because she hadn't received the government's letter yet, and had to listen to the news on the radio. After the story came out in the media, the phone lines were jammed for 6 hours. Patients were calling in, sobbing. One woman said, "How could you do this to us, Sally?" (Apparently the reason the British newspapers can write headlines like this is that people there actually talk that way.)

There were a few problems with the WHO study that were too complicated for newspaper stories.

First, the 95% confidence interval of that relative risk ranges between 1 and 4, observed Frankel. That means there could be a 4-fold increased risk--but there could also be no increased risk at all.

Second, these were case-control studies, which always have an inherent problem of sampling bias. The 3rd generation pills were thought to be safer, so high-risk women, from smoking and other causes, were put on the 3rd generation pills. "We're stacking the deck against 3rd generation," said Frankel.

Needle exchange in Vancouver

Now imagine a trip to the land of the Houyhnhms, or at least Canada, a land that is far from perfect but at least has universal health care, a land whose relatively rational inhabitants live in an uneasy relationship with their neighbors to the South, the Yahoos, or Congressmen.

"Needle exchange programs seem to stir more passion than any other issue in AIDS research," said Schechter, director of the Canadian HIV Trials Network. Needle exchanges are "widely accepted" throughout the world. The "notable exception is the U.S.," where they are banned in several states, including New Jersey, and where the federal government refuses to fund them.

In this story, said Schechter, "the major villain is going to be the government."

"Vancouver is a gritty downtown area," said Schechter. There were "stable rates" of HIV for several years. Then in 1995 they noticed an "explosive outbreak." They did cohort studies in several cities. HIV had an incidence of 18% a year--"the highest rate in the developed world."

Schechter published an article on the reasons for that outbreak in the journal AIDS, in 1997. "Needle exchange is not enough," he said. There was for example a lack of housing.

Schechter's studies stratified the population by more or less frequent users of the needle exchanges. Frequent users had a much higher prevalence of HIV. But they were very different populations. "Frequent users were much more likely to be using cocaine as the drug of choice. they were much more likely to be involved in the sex trade."

In the paper, Schechter wrote in the conclusions that this link between needle exchanges and HIV "should not be interpreted as causal," and that needle exchanges "attract higher risk intravenous drug users."

Simple, wrong

But some people didn't read the paper carefully. From this, said Schechter, "One can draw a very simplistic conclusion: Needle exchange programs must be spreading HIV."

After the study was published in AIDS, U.S. Congressman J. Dennis Hastert gave a press conference in which he said, "Vancouver studies have shown" that needle exchanges were spreading HIV. Hastert said, "I have a copy of the Vancouver study that shows that." Sen. Paul Coverdell proposed legislation to ban needle exchange programs forever.

There was an op-ed article in the New York Times, said Schechter. There were articles in the Washington Times, "which is a very strange newspaper, I gather."

Schechter could understand why the misunderstanding took off the way it did. "It fits well with existing ideology. It provides a nice sound or print bite."

As a doctor, Schechter recalled the phenomenon of perseveration, which is a symptom of the Korsakoff's psychosis of chronic alcholoics. It is an "inappropriate or unquestioning repetition of some behavior or speech." In the medical world, "it is often found in the medical chart," which sometimes grows to several inches in thickness, "the vehicle by which incorrect diagnoses are allowed to persist."

"In the media there is something called the wire service," said Schechter. This misinformation was "passed from newspaper to newspaper, with exactly the same sentences." He would get calls from public health workers all over the U.S., telling him, "Our local paper printed this story and now they want to ban needle exchanges."

Confounding factors

"The simple conclusion is wrong," said Schechter. "The not-so-simple conclusion is that needle exchange participants have a higher risk." Epidemiolgical confounding occurs when another factor is associated with the factor of interest.

One classic case of confounding is the correlation between alcohol and heart attacks, said Schechter. People who drink tend to have heart attacks. But that doesn't mean alcohol causes heart attacks. People who drink also tend to smoke cigarettes.

Schechter is realistic about the plight of the media. "Imagine a reporter going into his editor and saying, 'It's about confounding and we have to explain it to the reader.'"

"Three members of the Office of National Drug Control Policy visited us," said Schechter.

"When the people from the White House came, we spent an hour and a half trying to explain it to them," said Schechter. "People who use hospitals have higher death rates than people who don't. So hospitals must be killing people."

"They were clearly in the grips of an ideology," said Schechter. "It didn't make much difference what we said. Their report was already written."

###