The State of Drug Safety Systems

— Why FAERS continues to be the gold standard for regulators and industry.

MedpageToday

I've written a lot over the past few weeks about our policy efforts to better integrate adverse events data into the healthcare decision-making process. While the response has been overwhelmingly positive, we've also received a few questions and comments from people about the current state of the core adverse events dataset -- the FDA's Adverse Event Reporting System (FAERS).

Most of those questions fall into one of two categories:

1. Aren't there better/newer drug safety surveillance systems available now?

2. Doesn't FAERS have a number of limitations and biases that make analysis of those data difficult?

Are there new drug safety surveillance systems? Yes. Are they better than FAERS? Absolutely not.

Hundreds of millions of dollars have been spent by health regulators around the world (including the FDA's Sentinel project and its $150 million-plus cost) to develop active drug safety surveillance systems based on analyzing millions of electronic health records (EHR) "on demand." While in theory this "Big Data" approach seems to holds a lot of promise, the reality has proven to be a lot harder and colder.

A recent article in the journal Drug Safety by staff at the Institute for Safe Medications found that "the research programs and pilot systems created to study harms of licensed drugs proved largely unable to provide credible evidence of new, unsuspected drug adverse effects, and conflicting and contradictory results when seeking to confirm known harms. Major problems included a limited underlying terminology, few validation studies, and the need for additional statistical standards for these complex data."

In fact, the authors claimed that "six years after the building of Sentinel commenced it appears that the system has not yet been the primary data source in identifying a single new drug risk that led to a significant regulatory action such as a drug withdrawal, boxed warning, restriction, or contraindication."

The other future drug surveillance system may involve surveillance of social media, search engine terms, and other patient-generated content. While this all may seem to hold a lot of promise, it's very much in its infancy and not at all a reality today. It's anyone's guess if, one day in the future, it can grow into something tangible.

That means that when it comes to surveillance of drug safety issues in post-marketed drugs, FAERS has been, and will continue to be, the gold standard for regulators and industry.

So what about the second question -- the limitations and biases that may exist in FAERS? If you've followed our blog, you know that we've done a lot of work and published a lot of peer-reviewed research to debunk those notions of limitations and biases. I won't repeat all of those findings here, but invite you to read our work on the so-called Weber Effect and Stimulated Reporting in FAERS, as well as FAERS - Myths vs. Realities.

Instead, I thought it would be more instrumental to take a look at the most recent set of FAERS data we processed to get an understanding of exactly how much the value of FAERS has grown over the past few years.

Reporting into FAERS is Increasing Rapidly

The volume of reports into FAERS has increased significantly over the past 5 years. Clearly our efforts, as well as that of our partners, to publicize the importance of reporting and utilizing FAERS data has played some role in the increase.

Other factors include the fact that FDA has made it easier and quicker to directly report adverse events into FAERS and at the same time has cracked down on pharmaceutical companies for failing to report. Sometimes improving on what you have, instead of recreating the wheel (Sentinel) is the most expedient and cost effective solution.

Over the past 5 years, the number of primary suspect adverse event case reports has increased by 71%, from 525,000 in 2010 to 900,000 in 2014.

Underreporting in FAERS is Becoming Less of an Issue

One of the lingering concerns about FAERS data is the issue of underreporting. FAERS does not record every serious adverse event that occurs from a patient taking a prescription medication. However, the most widely used statistic about the rate of underreporting is now clearly outdated. A General Accounting Office (GAO) report published in 2000 claimed that 90% of serious adverse events went unreported in AERS (the predecessor system to FAERS). To provide some context, only 153,000 primary suspect cases were reported into AERS that year (that's 17% of the 2014 total).

While it's difficult to accurately calculate the current rate of underreporting, we believe it's more than fair to assume that 90% is now a gross overestimation. Data from [data broker] IMS Health revealed that in the same 2010 to 2014 period shown above, U.S. pharmaceutical sales increased by 19%, meaning that the rate of reporting of adverse events into FAERS has far outpaced the growth in prescription drug sales. Regardless of the true rate, FAERS still has the largest post-market collection of adverse events of any system.

Reporting of the Most Serious Adverse Events is Increasing Even More Rapidly

At the same time, the number of those primary suspect adverse event cases with the most serious outcomes (hospitalizations, deaths, and disabilities) has increased at any even faster rate. In 2010, those cases totaled 170,000 and by 2014 had grown to 331,000, an increase of almost 95%.

Hospitalization, Death, and Disability Outcomes from Adverse Events

This is particular important since the adverse events with the most serious outcomes are the ones that cost the healthcare system the most. The Agency for Healthcare Research and Quality (AHRQ) and the CDC estimate the average cost of a hospitalization from an adverse event at $10,033, the average cost of a death from an adverse event at $23,017, and the average cost of a disability from an adverse event at $11,471.

In total then, the most serious adverse events in 2014 linked by the highest level of possible causality to single drugs cost the healthcare system a reported $4.5 billion. And remember that this is only what's been reported. A reasonable (not 90%!) underreporting multiple should be factored into this equation.

Influences and Stimulated Reporting into FAERS Continues to be Marginalized

I think it's interesting to take a look at these reporting statistics from the 2014 year that clearly show how little potentially biased third parties are responsible for reporting of primary suspect cases. With healthcare providers and patients making up almost 99% of the reports filed in 2014, it's fair to say that the influence of litigation on these data is limited to none. And the benefit of FAERS is that they do request reporter type, as opposed to Internet searches that are unable to distinguish quality information from just plain trolling.

What to Expect in 2015?

In short, more of the same. Reporting to FAERS will continue to outpace the growth in prescription drug use, thereby further lessening the underreporting problem and strengthening the overall value of the dataset. Reporting of the most serious adverse events will continue to be the primary source of growth in reporting, and the influence of lawyers -- and other unrelated third parties -- will continue to be virtually extinct.

Regardless of naysayers, industry hype and hyperbole regarding future efforts, or the promise of Sentinel, FAERS is without peer; it's the most valuable asset we have in this country to monitor post market drug safety issues from prescription drugs. Improving on FAERS reporting and its data collection continues to be the fastest and most cost-effective way to monitor the safety of the U.S. population in relation to the use of an increasing amount of prescription drugs.

By successfully mining the available FAERS data and subsequently developing powerful analytic tools that reveal the true total cost of one drug versus another, as well as overturning a wide variety of overlooked safety concerns, AdverseEvents has made major strides in introducing these data and analytic tools to payers, providers, and pharmaceutical companies, all of whom are now using these data to help determine formulary decisions, cost projections, and safety alerts.

But there is a lot more that can be done. To better integrate the insight gleaned from this treasure trove of information into our healthcare system, we need FAERS to become de rigueur. Ideally, no hospital, doctor, pharmacist, insurer or health system should be prescribing or approving a drug without knowing the true safety profile of that specific medication over another. Similarly, no drug should be prescribed or approved without the patient knowing that safety profile and the downstream costs associated with taking that medication versus another.

Drug efficacy and safety go hand-in-hand to ensure the overall well-being of the U.S. population. Knowing whether a drug's side effects may cause hospitalization is as important as knowing whether the original symptoms may be remedied.

As a fellow drug safety advocate, you can play a role in making sure all safety aspects of prescription drugs are considered prior to their prescription. Let your voice be heard by taking the simple step of asking Centers for Medicare and Medicaid Services (CMS) to strengthen the Medicaid drug benefit within managed care programs, and by making regular reviews of FAERS data a component of improving patient safety and reducing systemic costs. More data, more analysis -- the safer the prescription drug market will be.

To let your voice be heard, contact CMS.

Brian Overstreet is co-founder and president of AdverseEvents, a California-based healthcare informatics company that improves patient safety and reduces systemic healthcare costs through the comprehensive analysis of postmarketing drug side effect data. This post originally appeared on the company's RxView blog.