Enhancing postmarketing surveillance: continuing challenges

In this issue, Zeitoun and colleagues 1 examined whether any relationship existed between the speed of European Medicines Agency (EMA) regulatory review of 161 medicines between 2001 and 2010 and reported postmarket safety events; they found no relevant correlation. Neither rapid EMA regulatory review nor approval near the deadline was associated with more reports of adverse events. The findings are reassuring, given the pressure that regulatory agencies face to achieve timely reviews and ensure public safety. There is evidence, however, that the rush to approve new medicines may compromise safety, at least in the United States. Since the approval process at the Food and Drug Administration (FDA) was accelerated through a legislative cost-shifting strategy known as the Prescription Drug User Fee Act in 1992, the numbers of both postmarketing black-box warning and market withdrawals have significantly increased 2. But the tensions affecting one regulatory system may not be applicable to another. The process of premarket approval should also be viewed as just one facet of pharmaceutical safety. While the findings of Zeitoun et al. 1 are encouraging, neither regulatory agencies nor the public should become complacent. Belated discovery of serious adverse drug events is not always the result of poorly conducted clinical trials, nefarious drug companies or weak regulatory oversight; some drug-related complications reflect the complexities of human biology. The most meticulous regulatory review will never guarantee the detection of all adverse drug effects. Maintaining a robust system of postmarketing surveillance must complement a thorough drug approval process. However, these processes too are often thwarted by limitations of health record data and corresponding decision support tools (e.g. dashboards, data cubes, etc.), and the practice of off-label prescribing.

Phosphodiesterase (PDE)-5 inhibitors for erectile dysfunction (ED) and non-arteritic ischemic optic neuropathy (NAION) provides a prime example of the elusive nature of drug safety. From 2002 through 2005, reports linking the use of phosphodiesterase PDE-5 inhibitors for ED and non-arteritic ischaemic optic neuropathy (NAION) appeared in the literature 3. Establishing the causal association between serious complications and drug use can be an arduous task, even when the adverse event is as clinically evident as vision loss. This complication had not been reported in premarketing clinical trials. Patients in the postmarketing phase described vision loss within 30 min to 36 h after drug use. A biologically plausible explanation for ischaemic injury to the optic nerve was evident from the start: PDE-5 inhibitors are capable of transiently lowering blood pressure. This transient dip in pressure could reduce vascular perfusion of the optic nerve head enough to result in localized infarction. By autumn 2005, Public Citizen, a nonprofit consumer organization, petitioned the FDA to add a black-box warning on all PDE-5 inhibitors, having gathered sufficient postmarketing evidence to consider the risk of vision loss a matter of public safety 4.

Skeptics noted that the men who use PDE-5 inhibitors for ED are also those who are at the highest risk for NAION – i.e. men who have diabetes and atherosclerotic vascular disease. In 2007, a large retrospective cohort study involving 4 157 357 male veterans found no increased risk of NAION in men prescribed PDE-5 inhibitors [odds ratio 1.02 (95% confidence interval 0.92–1.12)] 5. Similar findings using administrative data have continued to accumulate, despite anecdotal reports of the close temporal relationships between drug exposure and vision loss 6,7. One pharmaceutical company conducted its own prospective study and found a twofold increased risk of NAION within five half-lives of PDE-5 inhibitor use, a hazard that translates to three additional cases of NAION per 100 000 men 50 years of age and older each year 8. More than 13 years have passed since the first cases of PDE-5 inhibitor exposure and NAION were reported and the real risk of drug-related vision loss remains unsettled.

Rare but serious adverse drug-related events are a challenge to detect and even more daunting to establish as causal. An effective postmarketing surveillance system with clinically integrated health records and data systems is critical for proactively finding ‘rare events’ and assessing risks. The process of postmarketing surveillance in the United States suffers because the reporting of adverse events operates on a voluntary basis 9. Patients may traverse a myriad of different doctors and healthcare systems in a lifetime 10. Without clinical integration of electronic health records across institutions and between providers (e.g. lifetime electronic health record), regulatory agencies, doctors and researchers cannot be expected to identify rare but potentially serious adverse events in a meaningful time frame. Additionally, fully integrated healthcare records would require supporting software tools (data extraction tools, dashboards, data cubes, etc.) that would facilitate effective data management and analysis for postmarketing surveillance.

Why is postmarketing surveillance so important for public safety? One reason is that once a new medication is approved, its use in the medical community may vary considerably from the context in which it was tested for safety. The FDA approves drugs for indicated use but does not control how physicians prescribe medications. This allows physicians wide discretion to prescribe ‘off-label’ or outside of FDA indicated uses. Recent FDA guidance on physician off-label practices states that: ‘If physicians use a product for an indication not in the approved labeling, they have the responsibility to be well-informed about the product, to base its use on firm scientific rationale and on sound medical evidence, and to maintain records of the product’s use and effects’ 11. It is estimated that 20% of all prescriptions in the United States are currently prescribed outside of FDA indications for use, or ‘off-label’. There are no FDA requirements for a physician to inform patients that they are prescribing a medication ‘off-label’ 12. This practice has been a persistent obstacle to effective postmarketing surveillance of drugs because it can alter drug dosage, mode of administration and duration of use. It also means that drugs are being used in a population of patients that are likely to differ from those which safety was ascertained 12.

Drug safety requires a multifaceted approach from premarket approval to postmarketing surveillance. Rare but serious adverse events can be difficult to link causally with drug exposure, even when a plausible mechanism of injury exists. Optimal postmarketing surveillance cannot be based on voluntary reporting systems alone. The off-label use of drugs further confounds the problem of drug safety by exposing populations of patients that may fundamentally differ from those in premarket assessments of safety. Having an integrated large-scale medical record with a corresponding data retrieval system and decision support analytics for postmarketing surveillance would be an important first step in confronting these challenges.

Competing Interest

There are no competing interests to declare.

Funding/Support: Dr French is supported on an unrestricted grant from Research to Prevent Blindness, NY, NY and Department of Health and Human Services National Institutes of Health, NATIONAL EYE INSTITUTE Grant Number: 1R21EY024050-01A1. Drs Margo and Campbell have no financial disclosures to report.