After Approval, Into the Void: The Systemic Failure of Postmarket Drug Safety Monitoring

When the FDA grants approval to a new pharmaceutical agent, the decision is framed publicly as a determination that benefits outweigh risks. What is less frequently communicated—to clinicians, patients, or policymakers—is how provisional that determination actually is. Approval is granted on the basis of clinical trial data that, however rigorously collected, represents a controlled population observed over a finite period. The real test begins at market launch, when a drug enters the hands of millions of patients whose comorbidities, polypharmacy regimens, and demographic profiles were never fully represented in premarket studies. That real-world test, however, is monitored by a system that is, by most expert assessments, inadequately equipped to pass judgment.

The Architecture of a Flawed System

The United States postmarket surveillance framework rests primarily on the FDA Adverse Event Reporting System (FAERS), a voluntary database to which healthcare providers, manufacturers, and patients submit reports of suspected drug reactions. The operative word is voluntary. Unlike the mandatory pharmacovigilance requirements enforced in several European jurisdictions, the US model relies heavily on spontaneous reporting—a mechanism that the agency itself has acknowledged captures only a fraction of actual adverse events. Estimates in the peer-reviewed literature suggest that FAERS reflects somewhere between 1% and 10% of true adverse event incidence, a range so wide it renders population-level safety signal detection inherently unreliable.

Manufacturers bear a statutory obligation to submit reports of serious and unexpected adverse events within 15 days of becoming aware of them. Periodic safety update reports are also required. Yet enforcement of these requirements has historically been inconsistent. A 2021 analysis published in JAMA Internal Medicine identified thousands of delayed manufacturer submissions across multiple drug classes, with some reports filed years after the qualifying event. The FDA's capacity to audit submission completeness is constrained by resource limitations that have persisted across multiple administrations.

Financial Incentives Pointing in the Wrong Direction

Pharmacological development is extraordinarily expensive, and the return on investment is concentrated in the years immediately following patent protection. Within that commercial window, there is a structural disincentive for manufacturers to aggressively pursue safety signals that might generate label updates, restricted indications, or market withdrawal. Pharmacovigilance departments within pharmaceutical companies are cost centers, not revenue generators. While that framing is reductive, it reflects a documented tension between commercial imperatives and safety infrastructure investment.

Several researchers have noted that postmarket commitments—studies manufacturers agree to conduct as a condition of accelerated approval—are routinely delayed or never completed. A 2022 report from the Government Accountability Office found that a substantial proportion of postmarket study commitments remained open well beyond their agreed-upon completion dates, with limited regulatory consequence for sponsors who failed to deliver. The accelerated approval pathway, designed to expedite access to promising therapies for serious conditions, has in some cases functioned as a mechanism for indefinitely deferring the confirmatory evidence that was supposed to follow.

When Signals Go Unheard: Illustrative Cases

The consequences of surveillance failure are not theoretical. The saga of rofecoxib (Vioxx), withdrawn in 2004 after cardiovascular safety signals were belatedly acknowledged, remains the canonical illustration. Data suggesting elevated myocardial infarction risk had existed in the published literature years before withdrawal; the delay between signal emergence and regulatory action was estimated to have contributed to tens of thousands of preventable cardiac events in the United States alone. Subsequent congressional investigations revealed that internal FDA reviewers had raised concerns that were not acted upon with sufficient urgency.

More recently, the evolving safety profiles of certain fluoroquinolone antibiotics—particularly regarding peripheral neuropathy and aortic aneurysm risk—demonstrated how signals can persist in FAERS and case literature for extended periods before accumulating sufficient weight to prompt label revisions. Clinicians prescribing these agents during that interval had no formal mechanism through which emergent safety data was systematically communicated to them.

Sentinel and Its Limitations

In response to longstanding criticism of FAERS, the FDA launched the Sentinel System in 2008, a distributed network linking electronic health record and claims data from participating health systems to enable active, population-level safety surveillance. Sentinel represents a genuine methodological advance—it allows the agency to query real-world data on drug exposures and outcomes at a scale that spontaneous reporting cannot approach. By 2023, the network encompassed data on hundreds of millions of patient-years of observation.

Yet Sentinel is not a panacea. Its utility is constrained by the completeness and standardization of the underlying data sources, which vary considerably across participating institutions. Outcome misclassification, incomplete medication exposure records, and the absence of certain clinical variables limit the precision of signal detection. Additionally, Sentinel operates reactively—it is deployed to investigate hypotheses already under consideration rather than functioning as a continuous, hypothesis-generating surveillance instrument across all approved agents simultaneously. The system is a meaningful improvement over passive reporting, but it does not constitute comprehensive postmarket surveillance.

Structural Reforms Under Discussion

Among the reforms most frequently proposed in the clinical research literature, mandatory adverse event reporting for healthcare institutions stands out as a measure with significant potential impact. Countries including the United Kingdom have implemented systems in which hospital-level reporting is embedded in clinical workflow, substantially increasing signal capture rates. Whether such a mandate is politically feasible in the US context—given the administrative burden it would impose on already-strained health systems—remains debated.

Some researchers have advocated for expanded use of pragmatic postmarket trials embedded within large health systems, using randomization at the point of prescribing to generate comparative safety data in real-world populations. Others have proposed risk-stratified surveillance requirements, under which drugs approved via expedited pathways or with limited trial populations would be subject to more intensive postmarket monitoring obligations.

The FDA has taken incremental steps toward strengthening its pharmacovigilance framework, including expanded guidance on the use of real-world evidence for postmarket safety assessments. Whether these measures are commensurate with the scale of the problem is a question the research community has not resolved.

A Question of Institutional Priority

Ultimately, the postmarket surveillance gap reflects a broader question about where drug safety sits in the hierarchy of institutional priorities. Regulatory agencies face competing pressures—to accelerate approval timelines, to reduce industry burden, and to ensure that effective therapies reach patients without undue delay. Each of these pressures is legitimate. None of them is compatible with a surveillance infrastructure that remains, decades after its fundamental architecture was established, dependent in substantial part on voluntary reporting.

For clinicians managing patients on long-term pharmacotherapy, the practical implication is that the safety profile documented at the time of prescribing may not reflect the current state of evidence. Engaging with primary literature, monitoring regulatory communications, and maintaining a low threshold for adverse event reporting are not merely professional obligations—they are, given the limitations of the formal surveillance apparatus, necessary compensatory behaviors. The system, as currently constituted, requires clinician vigilance to fill gaps it should not have in the first place.