Post-marketing Surveillance of Prescription Drugs

What is post-marketing surveillance of prescription drugs? How is it done? What does it mean for clinicians?

As prescription drug abuse continues to be a problem, new technologies are being developed to track prescription drug abuse around the country. In this roundtable, clinicians and researchers will discuss post-marketing surveillance of prescription drugs, how it is done, and what the implications are for clinicians.

Stephen F. Butler, Ph.D:

Stephen F. Butler, Ph.D is the Senior Vice President and Chief Science Officer at Inflexxion, Inc. in Newton, MA.

Stephen Butler will answer the question what is post-marketing surveillance of prescription drugs?

Post-marketing surveillance of potentially addictive medications (generally Schedule II or III) began in earnest following the OxyContin outbreak in 1999 or 2000. In his book, Pain Killer, the New York Times reporter, Barry Meier (2003) documented this so-called "epidemic." The book portrays the frustration of providers, law enforcement officials, FDA officials, and the pharmaceutical company in early efforts to determine whether an outbreak was occurring and the role played by a particular product. At that time, no one had any data to which they could turn. Reports from physicians, addiction counselors, coroners, and police were sent to the FDA about an upsurge of abuse and deaths in local, often rural communities. These reports typically mentioned OxyContin as the problem. Up to that point, OxyContin, an extended release version of oxycodone, was thought to be less abusable, since the formulation's gradual release of the opioid was supposed to be associated with less "high" than the traditional versions (such as Percocet), where the all the opioid is released immediately. However, what appeared to be happening was that abusers could simply chew or crush the OxyContin and get an immediate release of 12 hours worth of active ingredient—leading to rapid addiction and overdoses.

In the early months of 2000, FDA officials were literally sifting through dueling anecdotes from local health and safety authorities, on the one hand, and those compiled by the pharmaceutical company on the other. The pharmaceutical company maintained that there was a general increase in opioid abuse, not just OxyContin. There was no particular problem with their product. Rather, the outbreak was due to a media frenzy. A later study, published in 2003 by the US Government Accountability Office (GAO), concluded that they could not assess the relationship between the increased availability of OxyContin and locations where it was abused and diverted because the data on abuse and diversion were not reliable, comprehensive, or timely. This lack of data resulted in ambiguity that not only fostered a lag in the authorities' ability to respond, but to this day prevents resolution as to whether the OxyContin "epidemic" was truly a product-specific epidemic or was merely a media construction.

Several groups, including a taskforce of the College on Problems of Drug Dependence (CPDD), recommended a series of actions, among which was establishing comprehensive post-marketing surveillance systems. This notion was based on the concept of public health disease surveillance. Public health surveillance has a long history of monitoring various diseases (especially infectious diseases) to determine whether and where an outbreak is occurring to give public health officials time to develop plans and to implement interventions. The Centers for Disease Control tracks 52 diseases and reports on these weekly. Following the anthrax attacks/scares that occurred in the wake of 9/11, larger and more systematic efforts were made to set up systems that can provide early detection of an outbreak of symptoms that might indicate a bioterrorism attack. Development and testing of these surveillance systems continues today.

Post-marketing surveillance of prescription medications builds on an analogy to disease surveillance and has adopted many of its designs and methods. That is, as a case of abuse, addiction, or death related to prescription opioids comes to the attention of some authority, that case is noted in some database and tracked to determine whether and where there is an increase in rates of abuse of a particular medication. In the case of medications, that can be an increase in a particular compound (such as hydrocodone or oxycodone) or a particular product (e.g., Vicodin or OxyContin). To be useful, post-marketing surveillance should have good geographic specificity and spread, and collect and analyze data in a "real-time" or "near real-time" fashion. As the GAO noted, databases on abuse rates that were in place in 2000 regularly took two years or more before data were ready to be analyzed—too late to be of much public health use. Modern surveillance systems are working out how to collect data that can be immediately analyzed and tracked, even on a daily basis.

Usually, surveillance involves determining if a target condition (in this instance, abuse cases) is increasing over a baseline rate or what would ordinarily be expected. For instance, if influenza is being tracked, the public health question is whether the number of cases this year is greater than would be expected in a normal year. In this way, the outbreak is defined in relation to some baseline occurrence of the disease. When the number of cases is significantly greater than the baseline, a "signal" is detected and reported to public health authorities. Likewise, abuse surveillance assumes that each community has a "baseline" level of abuse (since abuse of abusable substances does occur) and that a signal will entail a significant increase over this level of abuse. A signal could then lead to intervention. Obviously, and we have found this in our work, even the baseline abuse rates of some communities are extremely high. A surveillance system should be able to identify such high risk areas and track ups and downs in this baseline rate as well.

A major difference between surveillance of abuse and surveillance of disease is that, while cases of illness are usually detected in medical settings (hospitals, doctor offices, emergency departments), cases of abuse come into contact with authorities through a variety of points, including hospitals and emergency departments, but also admissions to substance abuse treatment, calls to poison control centers, or criminal justice agencies. Which of these provides an early warning is not yet clear, and systems have been developed to track data in all of these areas. One might ask about admissions to substance abuse treatment. If admission to treatment occurs at the end stage of an addiction, shouldn’t treatment admissions be a following indicator? Yet, many long-time addicts who enter treatment are connoisseurs of the various opioid products, and if a particular opioid product is increasing in availability and abuse in a local community, there is reason to believe that new admissions will report using it and allow a signal to be detected. What types of data should be monitored, that is, which data (hospital or emergency room admissions, arrests, substance abuse treatment, etc.) are leading indicators and which are following indictors remains an empirical question and an intense focus of current research.