For healthcare professionals only


When drugs don’t make it to market

BMJ 2015; 350 doi: (Published 09 June 2015) Cite this as: BMJ 2015;350:h2852
  1. Barbara Mintzes, senior lecturer
  1. 1Faculty of Pharmacy and Charles Perkins Centre, University of Sydney, Charles Perkins Centre D17, NSW Australia 2006
  1. Correspondence to: Barbara.mintzes{at}

We need to know why, and regulators should tell us

Drug regulation aims to protect public health by ensuring that only medicines that are effective and safe are marketed. The key regulatory innovation after the thalidomide tragedy was the requirement for systematic evidence of efficacy before marketing approval, with the understanding that no drug exposure is worth risking without evidence of benefit.

Over 50 years later, premarket evaluations remain much needed but imperfect, with efficacy standards not always ensuring clinical benefit for patients,1 and inadequate public communication on the available evidence and how it has been assessed. The United States Food and Drug Administration (FDA) is more open than many regulatory agencies, but, when it comes to non-approvals, the linked study by Lurie and colleagues (doi:10.1136/bmj.h2758) shows that commercial confidentiality continues to trump public health.2

If the FDA rejects an application, the letter to the manufacturer stating the reasons is confidential. Lurie and colleagues are FDA staff members who compared the content of FDA non-approval letters (n=61) with the firms’ press releases.2 This is the first published study to compare what regulators say in confidence to pharmaceutical manufacturers and what those manufacturers then say to the public. The differences are damning: 85% of the regulator’s safety concerns and 84% of efficacy concerns went unreported in manufacturers’ press releases, as did 96% of regulatory concerns about trial conduct or design. Increased death rates were not mentioned in six out of seven cases in which the FDA stated this as a reason for non-approval. In total, 87% of the FDA letters in this study cited efficacy or safety concerns as reasons for non-approval, and nearly half cited both.

Secrecy concerning non-approval of drugs has a similar potential to harm the public as trials with negative results that fail to see the light of day. Drugs are often approved in one country but not another. The fact that a drug is considered too ineffective, too unsafe, or too poorly tested to be marketed is highly relevant to treatment decisions elsewhere. In Canada, for example, the safety of the acne drug cyproterone-ethinylestradiol (Diane-35; Dianette) has been controversial,3 in part because of advertising to the public of questionable legality.4 This drug was not approved in the US. Knowledge of whether it was considered for approval, and, if so, why the FDA did not approve the drug, could help to inform the debate in Canada.

Many initially rejected drugs are eventually approved. Of 302 new drug applications in the US from 2000 to 2012, 151 (50%) were initially approved and 71 (24%) were approved after resubmissions.1 Some of the latter drugs became “blockbusters,” including gabapentin, pregabalin, duloxetine, aripiprazole, and rosuvastatin. Initial concerns over safety or efficacy might be only partially resolved during successful resubmission, and it is important that clinicians and the public are fully aware of all remaining issues as they might guide a more judicious approach to prescribing. Trials of unapproved drugs are also much less likely to be published than trials of licensed drugs, limiting the scientific knowledge that supports further drug development.5

Unlike the FDA, the European Medicines Agency (EMA) posts reasons for non-approval in European Public Assessment Reports (EPARs) on its website. A loophole that allowed non-disclosure if manufacturers withdrew applications before rejection was closed in 2012.6 International trade agreements are often cited as a reason that regulators must respect commercial confidentiality agreements. The EMA’s ability to publish the basis for decisions not to approve new drugs, however, strongly suggests that this argument need not prevent publication of the FDA’s non-approval letters.

Non-approval decisions account for just one of many differences between what the regulators know and what the public can see. Published trial reports of evidence submitted to regulators can also be highly misleading. High profile examples include the publication of interim results of the celecoxib CLASS trial as though these were the full clinical trial results,7 trials of antidepressants with negative results spun to look positive in publications,8 and a published pooled analysis of otherwise inaccessible unpublished oseltamivir trials, which misleadingly concluded that the drug prevented complications of flu.9 Internationally, regulators had full access to the data in each of these cases but did not correct omissions or inaccuracies in publicly available information.

Also missing from the public domain is the totality of evidence required to accurately judge a drug’s effectiveness and safety. Wieseler and colleagues compared 101 clinical study reports submitted to regulators with published articles on the same trials and found that many important outcomes were missing, including mortality, which was reported adequately in all clinical study reports measuring this outcome (n=92) but only 30% of corresponding articles or 53% of articles and clinical trial registry reports combined.10 How can readers interpret trial results without knowing how many people have died in each treatment arm?

The FDA’s judgments of trial misconduct are also inadequately reported in the published literature. If the FDA finds serious misconduct when it inspects a clinical trial site, data from that site are usually excluded from drug reviews. Among 71 published articles on 57 trials with serious infractions, however, 68 (96%) made no mention of the infraction and included no corrections, retractions, or omissions of data linked to infractions. In 39% of cases the infractions included uncorrected falsified results.11

What is the solution? Lurie and colleagues’ analysis points to two much needed steps.2 The FDA should follow the EMA’s lead and make non-approval decisions public. Commercial confidentiality has no place in the evaluation of the health effects of medicines. Secondly, although hampered by lack of disclosure of specific product and manufacturer names, this study is a welcome first step by regulators towards analysing and publishing data about discrepancies between the information they have available in pre-market and post-market submissions and reports that are in the public domain. Such engagement is welcome, must continue, and could contribute substantially to the quality and accuracy of information available to doctors and patients on the benefits and harms of medicines.


Cite this as: BMJ 2015;350:h2852


  • Research, doi:10.1136/bmj.h2758
  • Competing interests: I have read and understood the BMJ Group policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned; not externally peer reviewed.


View Abstract

Log in

Log in through your institution


* For online subscription