Brazen Scofflaws? Are Pharma Companies Really Completely Ignoring FDAAA?

Results reporting requirements are pretty clear. Maybe critics should re-check their methods?

Ben Goldacre has rather famously described the clinical trial reporting requirements in the Food and Drug Administration Amendments Act of 2007 as a “fake fix” that was being thoroughly “ignored” by the pharmaceutical industry.

Pharma: breaking the law in broad daylight?

He makes this sweeping, unconditional proclamation about the industry and its regulators on the basis of  a single study in the BMJ, blithely ignoring the fact that a) the authors of the study admitted that they could not adequately determine the number of studies that were meeting FDAAA requirements and b) a subsequent FDA review that identified only 15 trials potentially out of compliance, out of a pool of thousands.


Despite the fact that the FDA, which has access to more data, says that only a tiny fraction of studies are potentially noncompliant, Goldacre's frequently repeated claims that the law is being ignored seems to have caught on in the general run of journalistic and academic discussions about FDAAA.

And now there appears to be additional support for the idea that a large percentage of studies are noncompliant with FDAAA results reporting requirements, in the form of a new study in the Journal of Clinical Oncology: "Public Availability of Results of Trials Assessing Cancer Drugs in the United States" by Thi-Anh-Hoa Nguyen, et al.. In it, the authors report even lower levels of FDAAA compliance – a mere 20% of randomized clinical trials met requirements of posting results on clinicaltrials.gov within one year.

Unsurprisingly, the JCO results were immediately picked up and circulated uncritically by the usual suspects.

I have to admit not knowing much about pure academic and cooperative group trial operations, but I do know a lot about industry-run trials – simply put, I find the data as presented in the JCO study impossible to believe. Everyone I work with in pharma trials is painfully aware of the regulatory environment they work in. FDAAA compliance is a given, a no-brainer: large internal legal and compliance teams are everywhere, ensuring that the letter of the law is followed in clinical trial conduct. If anything, pharma sponsors are twitchily over-compliant with these kinds of regulations (for example, most still adhere to 100% verification of source documentation – sending monitors to physically examine every single record of every single enrolled patient - even after the FDA explicitly told them they didn't have to).

I realize that’s anecdotal evidence, but when such behavior is so pervasive, it’s difficult to buy into data that says it’s not happening at all. The idea that all pharmaceutical companies are ignoring a highly visible law that’s been on the books for 6 years is extraordinary. Are they really so brazenly breaking the rules? And is FDA abetting them by disseminating incorrect information?

Those are extraordinary claims, and would seem to require extraordinary evidence. The BMJ study had clear limitations that make its implications entirely unclear. Is the JCO article any better?

Some Issues

In fact, there appear to be at least two major issues that may have seriously compromised the JCO findings:

1. Studies that were certified as being eligible for delayed reporting requirements, but do not have their certification date listed.

The study authors make what I believe to be a completely unwarranted assumption:

In trials for approval of new drugs or approval for a new indication, a certification [permitting delayed results reporting] should be posted within 1 year and should be publicly available.

It’s unclear to me why the authors think the certifications “should be” publicly available. In re-reading FDAAA section 801, I don’t see any reference to that being a requirement. I suppose I could have missed it, but the authors provide a citation to a page that clearly does not list any such requirement.

But their methodology assumes that all trials that have a certification will have it posted:

If no results were posted at ClinicalTrials.gov, we determined whether the responsible party submitted a certification. In this case, we recorded the date of submission of the certification to ClinicalTrials.gov.

If a sponsor gets approval from FDA to delay reporting (as is routine for all drugs that are either not approved for any indication, or being studied for a new indication – i.e., the overwhelming majority of pharma drug trials), but doesn't post that approval on the registry, the JCO authors deem that trial “noncompliant”. This is not warranted: the company may have simply chosen not to post the certification despite being entirely FDAAA compliant.

2. Studies that were previously certified for delayed reporting and subsequently reported results

It is hard to tell how the authors treated this rather-substantial category of trials. If a trial was certified for delayed results reporting, but then subsequently published results, the certification date becomes difficult to find. Indeed, it appears in the case where there were results, the authors simply looked at the time from study completion to results posting. In effect, this would re-classify almost every single one of these trials from compliant to non-compliant. Consider this example trial:

• Phase 3 trial completes January 2010
• Certification of delayed results obtained December 2010 (compliant)
• FDA approval June 2013
• Results posted July 2013 (compliant)

In looking at the JCO paper's methods section, it really appears that this trial would be classified as reporting results 3.5 years after completion, and therefore be considered noncompliant with FDAAA. In fact, this trial is entirely kosher, and would be extremely typical for many phase 2 and 3 trials in industry.

Time for Some Data Transparency

The above two concerns may, in fact, be non-issues. They certainly appear to be implied in the JCO paper, but the wording isn't terribly detailed and could easily be giving me the wrong impression.

However, if either or both of these issues are real, they may affect the vast majority of "noncompliant" trials in this study. Given the fact that most clinical trials are either looking at new drugs, or looking at new indications for new drugs, these two issues may entirely explain the gap between the JCO study and the unequivocal FDA statements that contradict it.

I hope that, given the importance of transparency in research, the authors will be willing to post their data set publicly so that others can review their assumptions and independently verify their conclusions. It would be more than a bit ironic otherwise.

[Image credit: Shamless lawlessness via Flikr user willytronics.]

Thi-Anh-Hoa Nguyen, Agnes Dechartres, Soraya Belgherbi, and Philippe Ravaud (2013). Public Availability of Results of Trials Assessing Cancer Drugs in the United States JOURNAL OF CLINICAL ONCOLOGY DOI: 10.1200/JCO.2012.46.9577

Placebos: Banned in Helsinki?

One of the unintended consequences of my (admittedly, somewhat impulsive) decision to name this blog is that I get a fair bit of traffic from Google: people searching for placebo-related information.

Some recent searches have been about the proposed new revisions to the Declaration of Helsinki, and how the new draft version will prohibit or restrict the use of placebo controls in clinical trials. This was a bit puzzling, given that the publicly-released draft revisions [PDF] didn't appear to substantially change the DoH's placebo section.

Much of the confusion appears to be caused by a couple sources. First, the popular Pharmalot blog (whose approach to critical analysis I've noted before as being ... well ... occasionally unenthusiastic) covered it thus:

The draft, which was released earlier this week, is designed to update a version that was adopted in 2008 and many of the changes focus on the use of placebos. For instance, placebos are only permitted when no proven intervention exists; patients will not be subject to any risk or there must be ‘compelling and sound methodological reasons’ for using a placebo or less effective treatment.

This isn't a good summary of the changes, since the “for instance” items are for the most part slight re-wordings from the 2008 version, which itself didn't change much from the version adopted in 2000.

To see what I mean, take a look at the change-tracked version of the placebo section:

The benefits, risks, burdens and effectiveness of a new intervention must be tested against those of the best current proven intervention(s), except in the following circumstances: 

The use of placebo, or no treatment intervention is acceptable in studies where no current proven intervention exists; or 

Where for compelling and scientifically sound methodological reasons the use of any intervention less effective than the best proven one, placebo or no treatment is necessary to determine the efficacy or safety of an intervention 

and the patients who receive any intervention less effective than the best proven one, placebo or no treatment will not be subject to any additional risks of serious or irreversible harm as a result of not receiving the best proven intervention.  

Extreme care must be taken to avoid abuse of this option.

Really, there is only one significant change to this section: the strengthening of the existing reference to “best proven intervention” in the first sentence. It was already there, but has now been added to sentences 3 and 4. This is a reference to the use of active (non-placebo) comparators that are not the “best proven” intervention.

So, ironically, the biggest change to the placebo section is not about placebos at all.

This is a bit unfortunate, because to me it subtracts from the overall clarity of the section, since it's no longer exclusively about placebo despite still being titled “Use of Placebo”. The DoH has been consistently criticized during previous rounds of revision for becoming progressively less organized and coherently structured, and it certainly reads like a rambling list of semi-related thoughts – a classic “document by committee”. This lack of structure and clarity certainly hurt the DoH's effectiveness in shaping the world's approach to ethical clinical research.

Even worse, the revisions continue to leave unresolved the very real divisions that exist in ethical beliefs about placebo use in trials. The really dramatic revision to the placebo section happened over a decade ago, with the 2000 revision. Those changes, which introduced much of the strict wording in the current version, were extremely controversial, and resulted in the issuance of an extraordinary “Note of Clarification” that effectively softened the new and inflexible language. The 2008 version absorbed the wording from the Note of Clarification, and the resulting document is now vague enough that it is interpreted quite differently in different countries. (For more on the revision history and controversy, see this comprehensive review.)

The 2013 revision could have been an opportunity to try again to build a consensus around placebo use. At the very least, it could have acknowledged and clarified the division of beliefs on the topic. Instead, it sticks to its ambiguous phrasing which will continue to support multiple conflicting interpretations. This does not serve the ends of assuring the ethical conduct of clinical trials.

Ezekiel Emmanuel has been a long-time critic of the DoH's lack of clarity and structure. Earlier this month, he published a compact but forceful review of the ways in which the Declaration has become weakened by its long series of revisions:

Over the years problems with, and objections to, the document have accumulated. I propose that there are nine distinct problems with the current version of the Declaration of Helsinki: it has an incoherent structure; it confuses medical care and research; it addresses the wrong audience; it makes extraneous ethical provisions; it includes contradictions; it contains unnecessary repetitions; it uses multiple and poor phrasings; it includes excessive details; and it makes unjustified, unethical recommendations.

Importantly, Emmanuel also includes a proposed revision and restructuring of the DoH. In his version, much of the current wording around placebo use is retained, but it is absorbed into the larger concept of “Scientific Validity”, which adds important context to the decision about how to decide on a comparator arm in general.

Here is Emmanuel’s suggested revision:

Scientific Validity:  Research in biomedical and other sciences involving human participants must conform to generally accepted scientific principles, be based on a thorough knowledge of the scientific literature, other relevant sources of information, and suitable laboratory, and as necessary, animal experimentation.  Research must be conducted in a manner that will produce reliable and valid data.  To produce meaningful and valid data new interventions should be tested against the best current proven intervention. Sometimes it will be appropriate to test new interventions against placebo, or no treatment, when there is no current proven intervention or, where for compelling and scientifically sound methodological reasons the use of placebo is necessary to determine the efficacy and/or safety of an intervention and the patients who receive placebo, or no treatment, will not be subject to excessive risk or serious irreversible harm.  This option should not be abused.

Here, the scientific rationale for the use of placebo is placed in the greater context of selecting a control arm, which is itself subservient to the ethical imperative to only conduct studies that are scientifically valid. One can quibble with the wording (I still have issues with the use of “best proven” interventions, which I think is much too undefined here, as it is in the DoH, and glosses over some significant problems), but structurally this is a lot stronger, and provides firmer grounding for ethical decision making.

Emanuel, E. (2013). Reconsidering the Declaration of Helsinki The Lancet, 381 (9877), 1532-1533 DOI: 10.1016/S0140-6736(13)60970-8

[Image: Extra-strength chill pill, modified by the author, based on an original image by Flikr user mirjoran.]

Less than Jaw-Dropping: Half of Sites Are Below Average

Last week, the Tufts Center for the Study of Drug Development unleashed the latest in their occasional series of dire pronouncements about the state of pharmaceutical clinical trials.

One particular factoid from the CSDD "study" caught my attention:

Shocking performance stat:
57% of these racers won't medal!

* 11% of sites in a given trial typically fail to enroll a single patient, 37% under-enroll, 39% meet their enrollment targets, and 13% exceed their targets.

Many industry reporters uncritically recycled those numbers. Pharmalot noted:

Now, the bad news – 48 percent of the trial sites miss enrollment targets and study timelines often slip, causing extensions that are nearly double the original duration in order to meeting enrollment levels for all therapeutic areas.

(Fierce Biotech and Pharma Times also picked up the same themes and quotes from the Tufts PR.)

There are two serious problems with the data as reported.

One: no one – neither CSDD nor the journalists who loyally recycle its press releases – seem to remember this CSDD release from less than two years ago. It made the even-direr claim that

According to Tufts CSDD, two-thirds of investigative sites fail to meet the patient enrollment requirements for a given clinical trial.

If you believe both Tufts numbers, then it would appear that the number of under-performing sites has dropped almost 20% in just 20 months – from 67% in April 2011 to 48% in January 2013. For an industry as hidebound and slow-moving as drug development, this ought to be hailed as a startling and amazing improvement!

Maybe at the end of the day, 48% isn't a great number, but surely this would appear to indicate we're on the right track, right? Why would no one mention this?

Which leads me to problem two: I suspect that no one is connecting the 2 data points because no one is sure what it is we're even supposed to be measuring here.

In a clinical trial, a site's "enrollment target" is not an objectively-defined number. Different sponsors will have different ways of setting targets – in fact, the method for setting targets may vary from team to team within a single pharma company.

The simplest way to set a target is to divide the total number of expected patients by the number of sites. If you have 50 sites and want to enroll 500 patients, then viola ... everyone's got a "target" of 10 patients! But then as soon as some sites start exceeding their target, others will, by definition, fall short. That’s not necessarily a sign of underperformance – in fact, if a trial finishes enrollment dramatically ahead of schedule, there will almost certainly be a large number of "under target" sites.

Some sponsors and CROs get tricky about setting individual targets for each site. How do they set those? The short answer is: pretty arbitrarily. Targets are only partially based upon data from previous, similar (but not identical) trials, but are also shifted up or down by the (real or perceived) commercial urgency of the trial. They can also be influenced by a variety of subjective beliefs about the study protocol and an individual study manager's guesses about how the sites will perform.

If a trial ends with 0% of sites meeting their targets, the next trial in that indication will have a lower, more achievable target. The same will happen in the other direction: too-easy targets will be ratcheted up. The benchmark will jump around quite a bit over time.

As a result, "Percentage of trial sites meeting enrollment target" is, to put it bluntly, completely worthless as an aggregate performance metric. Not only will it change greatly based upon which set  of sponsors and studies you happen to look at, but even data from the same sponsors will wobble heavily over time.

Why does this matter?

There is a consensus that clinical development is much too slow -- we need to be striving to shorten clinical trial timelines and get drugs to market sooner. If we are going to make any headway in this effort, we need to accurately assess the forces that help or hinder the pace of development, and we absolutely must rigorously benchmark and test our work. The adoption of, and attention paid to unhelpful metrics will only confuse and delay our effort to improve the quality of speed of drug development.

[Photo of "underperforming" swimmers courtesy Boston Public Library on flikr.]

How Not to Report Clinical Trial Data: a Clear Example

I know it’s not even August yet, but I think we can close the nominations for "Worst Trial Metric of the Year".  The hands-down winner is Pharmalot, for the thoughtless publication of this article reviewing "Deaths During Clinical Trials" per year in India.  We’ll call it the Pharmalot Death Count, or PDC, and its easy to explain – it's just the total number of patients who died while enrolled in any clinical trial, regardless of cause, and reported as though it were an actual meaningful number.

(To make this even more execrable, Pharmalot actually calls this "Deaths attributed to clinical trials" in his opening sentence, although the actual data has exactly nothing to do with the attribution of the death.)

In fairness, Pharmalot is really only sharing the honors with a group of sensationalistic journalists in India who have jumped on these numbers.  But it has a much wider readership within the research community, and could have at least attempted to critically assess the data before repeating it (along with criticism from "experts").

The number of things wrong with this metric is a bit overwhelming.  I’m not even sure where to start.  Some of the obvious issues here:

1. No separation of trial-related versus non-trial-related.  Some effort is made to explain that there may be difficulty in determining whether a particular death was related to the study drug or not.  However, that obscures the fact that the PDC lumps together all deaths, whether they took an experimental medication or not. That means the PDC includes:

• Patients in control arms receiving standard of care and/or placebo, who died during the course of their trial.
• Patients whose deaths were entirely unrelated to their illness (eg, automobile accident victims)

2. No base rates.  When a raw death total is presented, a number of obvious questions should come to mind:  how many patients were in the trials?  How many deaths were there in patients with similar diseases who were not in trials?  The PDC doesn’t care about that kind of context

3. No sensitivity to trial design.  Many late-stage cancer clinical trials use Overall Survival (OS) as their primary endpoint – patients are literally in the trial until they die.  This isn’t considered unethical; it’s considered the gold standard of evidence in oncology.  If we ran shorter, less thorough trials, we could greatly reduce the PDC – would that be good for anyone?

Case Study: Zelboraf

FDA: "Highly effective, more personalized therapy"

PDC: "199 deaths attributed to Zelboraf trial!"

There is a fair body of evidence that participants in clinical trials fare about the same as (or possibly a bit better than) similar patients receiving standard of care therapy.  However, much of that evidence was accumulated in western countries: it is a fair question to ask if patients in India and other countries receive a similar benefit.  The PDC, however, adds nothing to our ability to answer that question.

So, for publicizing a metric that has zero utility, and using it to cast aspersions on the ethics of researchers, we congratulate Pharmalot and the PDC.