Bad Medicine
 
Why Data From Drug Companies May Be Hard To Swallow
 
By Gunjan Sinha

Lotronex (alosetron), manufactured by GlaxoSmithKline, appears to work much better than the placebo when percent change in pain severity is plotted against time (top graph).  The same data, replotted using the absolute values of pain severity by Public Citizen (bottom graph) shows that, in actuality, Lotronex had almost a negligible effect in pain reduction.  Both graphs appeared in the Lancet.   When the actual effect of a cause is small, plotting the results as a percent of the placebo, or control, value always makes the effect look much larger than it really is.  When the actual, absolute values are plotted alongside the placebo value(s), it can readily be seen how much of an effect there actually is (compare the top [percent] and bottom [absolute] graphs).     Some scientists argue that the FDA has become so chummy with industry partly because of the Prescription Drug User Fee Act, passed in 1992.  The act requires firms to pay the FDA almost $500,000 in total fees for each approved drug. Such fees account for almost half the agency’s cost of reviewing drugs.   "I was shocked the FDA buckled even after they’d seen the obfuscation and the attempts to hide data.  They seemed more comfortable working with the company than with their own staff," Stolley grouses.  Lotronex is now back on the market.  But only authorized doctors can prescribe it, and patients must sign an agreement stating that they fully understand the hazards.  Here's hoping that for them, it is truly a risk worth taking.

Getting drugs on the market means playing games. So says Peter Lurie of The Health Research Group (Public Citizen), an interest group founded by Ralph Nader and based in Washington, D.C. Of course, it's the agency's mission to be leery. But lately pharmaceutical companies are giving groups like Lurie's more to be leery about. Drug firms now wield a great deal of control over their research, Lurie charges, and they are frequently manipulating their data or withholding unfavorable results entirely.
 
One of Public Citizen's latest battles is over a drug for irritable bowel syndrome (IBS). Three years ago the Food and Drug Administration approved Lotronex (alosetron hydrochloride), the first agent to treat the disorder specifically. As published in the Lancet, clinical trials in women revealed that 41 percent taking the drug felt some relief, as did 29 percent taking a placebo.
 
The data, Lurie insists, "are incredibly misleading." One figure, for example, plots percentage change on one axis and time on the other. There are two problems with this. First, plotting percentage change instead of absolute change makes the effectiveness of the drug appear large. Second, the graph omits data from the first month, during which the drug and placebo worked almost identically.
 
Public Citizen replotted the data using absolute values. The graph, which the Lancet published in a letter, better represents the drug's "marginal" efficacy, the group argues.
 
"I don't understand the accusation," responds Michael Camilleri of the Mayo Clinic in Rochester, Minn., who led the study. "Such presentation is standard and accepted in peer-reviewed scientific journals. The data clearly show that the drug was better than the placebo for months two and three."
 
But many observers believe that drug companies go too far.
 
"It shouldn't happen in the scientific literature," insists Bob Goodman, founder of New York City-based No Free Lunch, which is focused on reining in the marketing ploys of drug companies. "Doctors should be able to decide the appropriateness of a drug. But how can they when drug companies leave out crucial information?"
 
Goodman is referring to another common practice: excluding data.
 
Illustrating the point is the ongoing controversy over Cox-2 inhibitors, touted as a safer alternative to nonsteroidal anti-inflammatories such as ibuprofen. Sales of one, Celebrex, reached a whopping $3 billion (U.S.) in 2001.
 
But last year, the Washington Post revealed that Pharmacia, the drug's maker, had published just six months of results. Data for the next six months indicated that patients on Celebrex suffered complications such as stomach ulcers at the same rate as those taking older medications. This information became public only because one of the paper's reviewers happened to be on the drug's FDA review committee.
 
Pharmacia says that the data for the last six months were too flawed to include them.
 
There are also rumblings that even though the FDA is aware of such practices, the agency is increasingly acting more favourably toward drug companies. The Lotronex story again provides the spark for that charge.
 
After the drug hit the market in February, 2000, the FDA assigned Paul Stolley of the University of Maryland to review the drug's side-effects profile. Stolley noticed a distressing pattern. Day after day he would see reports of patients being hospitalized, presumably because of Lotronex.
 
"And this for a disease that never leads to hospitalization, never perforates your colon, and is not life-threatening," Stolley says.
 
GlaxoSmithKline, the drug's maker, pulled Lotronex off the shelves in 2000 after 49 reports of ischemic colitis and three deaths. A few months later, responding in part to requests from IBS advocacy groups, the company appealed to the FDA to bring the drug back.
 
That move alarmed Stolley, who felt that the risks far outweighed the drug's marginal benefit. But when he spoke up, he was shut out. "FDA personnel were told not to discuss the case with me," says Stolley, who had consulted for the FDA for the past 30 years.
 
Others were opposed to the drug, but "they were intimidated," says Stolley, who now works for Public Citizen.
 
Some scientists argue that the FDA has become so chummy with the drug industry partly because of the Prescription Drug User Fee Act, passed in 1992. The act requires firms to pay the FDA almost $500,000 in total fees for each approved drug. Such fees account for almost half the agency's cost of reviewing drugs.
 
"I was shocked the FDA buckled even after they'd seen the obfuscation and the attempts to hide data," Stolley says. "They seemed more comfortable working with the company than with their own staff."
 
Lotronex is now back on the market. But only authorized doctors can prescribe it, and patients must sign an agreement stating that they fully understand the hazards.
 
Academic researchers who carry out drug investigations may not always be aware of data manipulation. A recent study in the New England Journal Of Medicine surveyed 108 medical schools in the U.S. and found that only 1 per cent of contracts between industry and academic institutions required that every researcher of a multi-centre study have access to all data.
 
And less than 1 per cent of contracts guaranteed that results would be published at all, ensuring that negative results are not published.

Gunjan Sinha is based in Frankfort, Germany.