Medical Examiner

Rent-a-Researcher

Did a British university sell out to Procter & Gamble?

Earlier this month, Sheffield University in Britain offered $252,000 to one of its senior medical professors, Aubrey Blumsohn. According to a copy of a proposed settlement released by Blumsohn, the university promised to pay him if he would agree to leave his post and not make “any detrimental or derogatory statements” about Sheffield or its employees. For several years, Blumsohn had been complaining of scientific misconduct. His concerns primarily revolved around a $250,000 research contract between Sheffield and the Ohio-based Procter & Gamble Pharmaceuticals. Blumsohn claimed that the company had denied him access to key data and then tried to ghostwrite his analysis of it. He further alleged that P&G had engaged in such practices before.

Why did Sheffield, a top-flight research university, try to silence and get rid of Blumsohn? The answer appears to lie in the complex and increasingly compromised relationships that have grown up between some research universities and the pharmaceutical industry. In 2001, the editors of nearly a dozen prominent medical journals warned that growing industry interference with academic research (from study design to data analysis and publication) was threatening the objectivity and trustworthiness of medical research. The editors issued new guidelines requiring all authors publishing in the journals to verify that they “had full access to all of the data” related to their studies and that they took “complete responsibility” for “the accuracy of the data analysis.”

But in the years since, universities with medical schools have become dependent on drug companies for an ever-larger share of their research budgets—roughly 80 percent of clinical research is now privately funded. And drug companies, in turn, have pressed for greater control over the research process, making it easier for them to obscure or delete negative results from published academic papers. Earlier this month, the New England Journal of Medicine accused Merck of failing to report three patient deaths in the trial that led to FDA approval of the painkiller Vioxx, which was pulled from the market last year because of its association with heart attacks and strokes. The careful record Blumsohn kept of his dealings with Procter & Gamble and Sheffield suggests that P&G didn’t control academic research on its own. It needed Sheffield University to permit incursions on scholarly independence.

In the summer of 2002, Blumsohn, a senior lecturer and bone metabolism specialist, and Dr. Richard Eastell, Sheffield’s research dean, signed a $250,000 research contract with Procter & Gamble. Blumsohn and Eastell were to evaluate the effectiveness of P&G’s osteoporosis drug, Actonel. The goal was not to win FDA approval; Actonel was already being widely prescribed. Instead, the Sheffield study would shed further light on how Actonel affects women’s bones and their susceptibility to fractures. According to Blumsohn, Eastell had already reviewed blood and urine samples from two previous P&G clinical trials of Actonel. Now Blumsohn was supposed to evaluate a third trial, with the aim of providing a final analysis of all three.

But in the past, it seemed, P&G had not allowed Eastell to perform his own data analysis. In an e-mail that Eastell wrote to P&G and copied Blumsohn on, he confessed that while presenting a paper at the International Osteoporosis Foundation, he had been unable to respond to questions about his own research posed by a fellow academic. “I think that to avoid criticism in the future it would be good if we could say that we had done the analyses independently,” Eastell wrote in the e-mail. He suggested that Blumsohn be entrusted with the independent analysis, so he could vouch for results that would be published under both their names.

Blumsohn and his staff reviewed thousands of blood and urine samples from women with osteoporosis. At this stage, they were “blinded” from knowing which patients had taken Actonel and which had taken a placebo. This helped to ensure objectivity. But when he finished examining the samples in December 2002, Blumsohn says he asked P&G to release the codes for the raw data so he could independently interpret the results.

Blumsohn requested the data access codes for 18 months, as numerous e-mails and other records document (here’s one). P&G officials wrote back refusing to permit independent access to the data. However, in a written statement, the company denied that it withheld necessary data. “We have appropriately shared our clinical data with both investigators and regulatory authorities, and have conducted our business with the highest of standards.”

Meanwhile, Blumsohn says P&G began to analyze his Actonel data and write up the final results for him to present at the American Society of Bone and Mineral Research in Minneapolis in fall 2003. The previous April, P&G statistician Ian Barton informed Blumsohn and Eastell by e-mail that Mary Royer, the company’s medical ghostwriter, would help write up the Actonel manuscripts for publication. Blumsohn and Eastell would both be listed as authors. Barton emphasized that the ghostwriter was “familiar with … our key messages.”

By now, Blumsohn thought he knew what those “key messages” were. In 2004, P&G’s main rival, Merck, was due to publish a head-to-head study comparing its osteoporosis drug, Fosamax, with Actonel. Many doctors considered Fosamax more effective at increasing bone density and decreasing the rate at which bones degenerate—and thus probably also more effective in preventing fractures, the biggest concern for women with osteoporosis. Fosamax’s global sales were $3 billion a year, compared to about $1 billion for Actonel. In the summer of 2003, Blumsohn received a copy of P&G’s proposed “statistical plan” for analyzing his data. It stated that the purpose of his research was to bring about an “Osteoporosis Paradigm Shift.”

Eastell’s earlier research asserting P&G’s claims about the effectiveness of Actonel appeared in June in the prestigious Journal of Bone and Mineral Research. Eastell and his co-investigators stated that “all authors had full access to the data and analyses.” Based on Eastell’s earlier e-mail, Blumsohn knew that wasn’t true and that Eastell had most likely violated the new safeguards that medical journal editors had put in place in 2001. Blumsohn says he warned Eastell they could both be accused of scientific fraud if they kept authoring papers without seeing the underlying data. A few days later, P&G’s Barton sent an e-mail reiterating that Blumsohn could not perform his own independent analysis of his data but could come to P&G’s offices to look at it.

When Blumsohn sat down with Barton at the company’s Surrey headquarters in late July, he says he spotted something peculiar. In one critical graph showing how Actonel affects fracture rates, Blumsohn noticed that 40 percent of the patient data was missing. Inclusion of the data, he thought, would have disproved P&G’s “key message” about Actonel’s effectiveness in reducing bone fractures. Several months later, Blumsohn recorded a meeting in which Barton expressed concern that if P&G included the missing 40 percent of the data, Merck would exploit the results. “Because that is contradicting our original manuscript,” he said. “I just know what Merck are like. I think they are going to use it.”

P&G denies that it skewed data to achieve desired results, saying that Blumsohn “was given access to all of the data related to his research.” But the company’s previous written statements seem to contradict this assertion. When Blumsohn’s lawyer filed a formal data request on his behalf, P&G responded: “It is not the standard practice of P&G to allow unlimited access to raw data from clinical trials to individual investigators, as these data are proprietary.”

In November, Blumsohn won a few concessions. He says P&G agreed to remove the graph he’d objected to from an oral presentation and to delete some text from a paper appearing in his name. But P&G’s educational materials and other writings continued to make assertions about Actonel’s effectiveness, which Blumsohn believed the data he’d seen did not support.

Increasingly, Blumsohn felt he was doing battle not only with P&G but with his university. Shortly after Blumsohn complained about the apparent manipulation of his Actonel data, he recorded a conversation in which Eastell warned, “The only thing we have to watch all the time is our relationship with P&G.” The P&G money “is a good source of income, we have got to really watch it.”

Over the next 22 months, Blumsohn wrote formal complaint letters to various Sheffield officials. The university didn’t investigate his claims, according to an article in the British Times Higher Education Supplement. In July, Blumsohn announced that he would go public with his concerns. Sheffield suspended him in September, citing, among other things, his “refusal to comply with a reasonable management instruction by briefing journalists.” Sheffield then offered him the $300,000. Sheffield states that it repeatedly asked Blumsohn to provide evidence supporting his allegations and urged him to bring that evidence “through the proper channels.” The university says legal negotiations were initiated “at Blumsohn’s request” and “undertaken in good faith by the University.”

Universities have long accepted funding from pharmaceutical companies to conduct clinical drug trials. But in the past, their professors insisted on running those trials independently of the sponsor. As the Blumsohn case makes clear, this arm’s-length relationship appears to be breaking down. Earlier this month, the Wall Street Journal reported on the growing willingness of some academics to sign their names—and lend their prestige—to articles and editorials penned by drug-company ghostwriters. In addition to the Vioxx episode, recent reports indicate that published academic studies related to the drugs Celebrex, Paxil, and Zoloft appear to have been skewed when their authors permitted the suppression of negative results.

The British Parliament has promised to investigate Blumsohn’s allegations. And in January, medical journal editors will be gathering once again to discuss what can be done to restore the integrity of the research they publish. One idea that’s been floated in the United States is a new arm of the National Institutes of Health that could serve as a repository for complete clinical trial data while also monitoring trials and verifying the accuracy of published results. Whatever the solution, something needs to be done soon. Scholarly independence has already taken too many hits.