• Why the world's most treasured paintings are turning black • Mathematicians working with NSA have 'little sense of social responsibility' • A way to clean fusion energy at last?

Drug companies, regulatory bodies and research groups are in a deadlock over plans to throw open the vaults of sensitive data gleaned from clinical trials. But just as the battle seems to be coming to a head, a study (B. Wieseler et al. PLoS Medicine 10, e1001526; 2013) has revealed exactly why the restricted information could be so valuable to researchers.

According to the analysis, crucial trial information, such as mortality rates and serious side effects, is missing from much published data. But it can frequently be found in standard non-public documents prepared by industry, known as clinical study reports (CSRs). Missing information uncovered by the study includes details of depression symptoms in trials for antidepressant drugs, and details of heart attacks and strokes in diabetes-drug trials. “These are very, very important variables and outcomes,” says the study’s lead author, Beate Wieseler, head of the drug-assessment department at the Institute for Quality and Efficiency in Health Care in Cologne, Germany.

Wieseler strongly supports making all CSRs publicly available. “That is not an option, but a necessity,” she says. “There should be no question that these documents be made available.”

Credit: Source: B. Wieseler et al. PLoS Medicine 10, e1001526; 2013

The issue of CSR access is shaping up to be a key sticking point in a major European push to make available more data from clinical trials. Drug companies send the reports to the European Union’s European Medicines Agency (EMA) when they apply for licences to sell their products, and the agency has made clear that it would like to release the forms as part of a drive to increase transparency. It is currently drawing up a policy to that effect. Advocates of transparency say that such a policy will allow greater scrutiny of the benefits and harms of drugs, and will help researchers who are working on cures for diseases.

Some companies, including pharmaceuticals giants Roche, based in Basel, Switzerland, and GlaxoSmithKline in London, have already said that they will make their CSRs available to vetted researchers. But industry has pushed back against the wider moves to greater transparency.

In an e-mail to Nature, Richard Bergström, the director-general of the European Federation of Pharmaceutical Industries and Associations (EFPIA), said that in their current state, CSRs are fundamentally unsuitable for publication. The EFPIA believes that the EMA’s proposed mechanisms for protecting the commercially sensitive and personal details contained in many of these documents are inadequate — the mechanisms require that information be ‘de-identified’ and released only for bona fide research purposes. Such personal data must be properly redacted if these documents are to be made public, says Bergström.

“My members are very concerned about this,” he says. “If the EMA accepts our redactions, we have no problem.” If the agency disregards the EFPIA’s concerns, however, Bergström warns that there may be a series of lawsuits against the EMA.

The EMA has already been taken to court by two biotechnology companies, AbbVie of North Chicago, Illinois, and InterMune of Brisbane, California, to block release of their information under existing rules, whereby any researcher can request information. The cases are ongoing and have severely restricted releases of other requested data.

But advocates of greater transparency are fighting for the EMA to release all data, which they say can easily be anonymized.

In the study, Wieseler and her team examined the information in 101 CSRs given to them by pharmaceutical companies, and compared it with the information about the same clinical trials that is available in the public domain, found in journal publications and reports in trial registries.

What’s portrayed in the journal world does not reflect the whole truth.

The unpublished sources provided considerably more information on the key findings, or outcomes, relating to the treatments being assessed, including mortality rates and adverse events. In total, the team found 1,080 outcomes relevant to patients. Complete information was available for 86% of these in the CSRs, but for only 39% in the publicly available information. When looking only at outcomes related to harm — such as adverse reactions to a drug — the researchers found that complete information was found for 87% of outcomes in the CSRs, but for only 43% in public records. In some cases, no public information was available at all for a given CSR. But when looking only at cases in which there was both a CSR and a public document, the researchers found similar proportions of missing information in the publicly available records (see ‘Inside drug-company data vaults’).

“Does it surprise me? No. Is this stuff really important? Yes,” says Carl Heneghan, director of the Centre for Evidence-Based Medicine at the University of Oxford, UK, and co-founder of a group called AllTrials, which campaigns for more clinical-trial transparency. “It’s becoming obvious that what’s portrayed in the journal world does not reflect the whole truth.”

In a public consultation that ended last week, the EMA’s proposals attracted support from other quarters. The Medicines and Healthcare Products Regulatory Agency, the UK drug regulator, says that it welcomes the EMA’s proposals, “and will work to align our position with theirs”.

A joint response from UK biomedical-research funders, including the Wellcome Trust and the Medical Research Council also backs the EMA’s plans. However, the group voices concerns over the sharing of data from individual patients; it says that more safeguards should be put in place to ensure that data are given only to trusted researchers who will not “wrongfully contradict” the results of trials.

The EMA hopes to have its policy in place by the start of next year. Documents without sensitive personal data will be made available for download from its website, whereas those that do contain such data will be available only after anonymization to vetted researchers. Documents with commercially confidential information will be released separately.