In July, the Journal of the American Medical Association printed a Research Letter and an on-line editorial on pharmaceutical companies’ late expedited (15 day) reports to the FDA. The article (behind a paywall) can be found here. and the editorial here and here. The study is from the University of Minnesota. The findings were based on data available to the public from the FDA quarterly post-marketing FAERS safety data files from 2004 to 2014.
In a nutshell, the authors looked at over 1,600,000 reports and found that over 160,000 (10%) were not received by the FDA within the 15-calendar day time frame required by federal regulations. Apparently these are only initial reports with follow-ups not included. Over 40,000 involved patient deaths. Three percent of the reports were three to six months late, and 3 percent were over 6 months late. Adjusted rates of reports taking 16–90 days to reach FDA were 6.42% for events with patient death and 5.19% for events without patient death.
This received some press coverage in late July with a handful of stories in newspapers and on social media but it does not seem to have gone viral. See for example, this article and this article.
Although the methodology may be criticized for some probably minor issues, the overall data and conclusions appear to be consistent with anecdotal reports and FDA inspection reports and Warning Letters. Thus the article itself makes some very good points.
However, the editorial in JAMA is another story being a bit naïve and not aware or informed of how the PV system works in the US. For example, here are some statements from the editorial:
- “Such reporting delays should never occur, as they mean that more patients are exposed to potentially avoidable serious harm, including death.”The first part of the statement is true (they should never occur) but the second is something of a presumption. I am not aware of any data supporting the contention that delays in reporting of ICSRs can cause harm and death though it does certainly seem to be possible. The reason for my skepticism is that many post-marketing reports are from consumers or are not medically validated or have insufficient detail to produce an immediate response by the company or FDA in terms of stopping sales, initiating a recall or withdrawal etc. Unlike clinical trial SUSARs, spontaneous reports are less consistently reliable. Some clinical trials will be stopped or altered to minimize risk based on a handful of cases (or even a single case). But that is a different matter from the post-marketing situation. Nevertheless, this statement is more or less true: data should be sent to FDA in the legally required time frame to minimize patient risk.What we don’t know for these reports is why the reports were late. That is the critical issue. In the world of Quality Systems, it is almost impossible to correct a problem unless one understands the cause(s) of the problem(s).
- Are the late reports still coming from large pharma companies who, more or less, got their acts together and set up more solid ICSR expedited reporting over the last decade?
- Are more reports coming from small companies, generic houses or others with fewer resources and money dedicated to safety reporting?
- Are clinical research organizations (CROs) involved? How do their functions compare with pharma companies doing the reporting themselves? That is, does out-sourcing play a role here.
- What percent of late cases were from abroad vs those from the US?
- For each of these sources of late reports, what is the trend over this ten-year period studied? Is the problem static, worsening or improving?
- For those companies that were penalized or cited by FDA for late reporting (e.g. issued Warning Letters or put in place a CAPA), was there an improvement in timely reporting?
- Did these reports involve newly approved drugs or old (off patent, generic etc.) drugs? The lateness is more critical for newly approved drugs with a still developing safety profile rather than old drugs where the safety profile is better characterized.
According to the news reports found on-line, FDA did not comment on this report when asked. One imagines they will at some point and hopefully will address some of these questions.
- “However, no disciplinary actions have been taken when companies fail to submit reports to the FDA in the time frame required. Clearly, the lack of consequences contributes to a lack of deterrence for these illegal and dangerous delays.”
This is not the case. The issue hinges on the definition of “disciplinary actions”. FDA has issued Warning Letters and 483/EIRs which note late reporting and require a response within 15 days and often the implementation of corrective actions (CAPAs). Whether there is a lack of deterrence is hard to determine based on this article; it is a separate issue.
- “There is another enforcement tool that the FDA could begin to deploy immediately: suspending drug sales or withdrawing drug approval.”This is certainly true but I think the author of the editorial has not really thought this through. Suspending sales or withdrawing the NDA will surely harm the pharmaceutical company but will, most likely, harm the patients, public and prescribers even more. If the drug has no alternatives or generic equivalents then the public is being deprived of the drug not because a new safety problem and risk to patients have been found but simply to punish a corporation. That is obviously not sound public policy.There are many other ways to punish companies for late reports without harming the end users. Most other punishments or incentives to rigorously comply with reporting deadlines revolve around money. Perhaps a more effective way would be for the FDA to penalize the company $1000 for each day a case is late. Five days late would cost $5000. A three month late case would produce a penalty of about $90,000. Or make a sliding scale, increase the penalty by $1000 for each additional day late: 1 day late: $1000, 2 days late: $1000 + $2000 = $3000; 5 days late: $15,000 etc. This mounts up quickly and would be a MAJOR INDUCEMENT for companies to put in place processes to minimize late cases. Sometimes, though, there are legitimate reasons for lateness beyond the control of the company and some mechanism would need to be put in place to make exceptions to the draconian monetary punishment. This could all be worked out and put in place relatively easily.
- “One improvement would be for AE reports to go directly to the FDA instead of via the manufacturer. This would address the delay problem…”But again naïve on the operational level. It would not address the delay problem fully as many reports come from outside the US and there is simply no way that foreign reporters would send information to the US FDA. Many countries require domestic cases to be reported locally and in the local language. The reporters simply would not send the reports (perhaps translated into English) to FDA also. These cases are then picked up locally by the pharma company and sent to FDA and other health agencies in the 15 day timeframe.Although the US has been the major supplier of safety reports, the percentage from the US is dropping as other countries develop their own PV reporting systems. As of May 2015, only 48% of cases in the Uppsala Monitoring Centre’s database (Vigibase) were from the US with the remainder from the rest of the world. The percent of cases coming from the US is dropping. So presumably a significant percentage of the cases sent to FDA (both on time and late) were from abroad and this percentage will likely increase. Will FDA seek these reports out from other agencies or will pharma companies still have to report ex-US cases to FDA? Thus there doesn’t seem to be a practical way to remove pharma companies from the reporting system.Another issue would be that having FDA receive all (or some) reports would shift the “reporting burden” to FDA away from the companies. The mechanisms put in place by companies to handle these cases and to prepare other safety reports (e.g. PSURs) for FDA and other health agencies is in the range of tens to hundreds of millions of dollars. There is no published data that I am aware of on these costs other than reports (cited in an earlier Bart’s Corner on the Business of Medicine) noting that pharmacovigilance costs are already in the billions of dollars and will rise to some $6 billion per year by the end of the decade.FDA would now have to set up systems to make sure that the reports are handled rapidly and in large volumes: immediate reviewed, triaged and entered into the database (FAERS), E2B or MedWatch forms prepared, follow-up data sought and entered into the database and, hopefully, transmitted to the pharma companies who sell the drug. This is not trivial and would cost tens to hundreds of millions of dollars to set up to handle the hundreds of thousands of reports received each year.Perhaps much of this will become moot when the nation moves to electronic health records whereby many or most adverse events can be “pulled” by FDA rather than relying on active reporting by patients, health care practitioners and pharma companies. One doubts we will see this in our lifetime.
- “Physicians and their patients must be knowledgeable of benefits, harms and alternatives for a wide choice of treatments, especially those recently approved for which clinical experience is limited.”Absolutely true. No one could possibly disagree with this statement. What is not clear is whether making the significant efforts to decrease late reporting from 10% to, say, 2% will indeed make a difference to the public health. One would think so but the question is quantitative and not answerable now with data. Should we divert limited money and resources to this problem versus spending it elsewhere? Should it only involve death reports? Should it exclude old drugs (e.g. on the market more than 10 years)? What is an acceptable late level? Obviously, no one in government will ever want to say publicly that we’ll accept a 2% late report level (or 3 airplane crashes a year). The goal is always zero percent late. But the practicality is that getting from 98% to 99.9% is often far harder and expensive than getting from 90% to 98%. Should FDA prioritize this over other pressing issues they are dealing with? These are all policy questions and to address them we need more data.
In summary, this paper provides interesting and important information that, as always, produces more questions that need answering if we are to improve this situation. In the meanwhile, on a practical level, pharma companies and CROs should have a look at their on-time/late expedited reporting data and put in place procedures to minimize delays with the goal of no late reports. And, obviously, if you’re not tracking late reports, start to do so immediately!