Why the GLP-1 Era Requires a New Approach to Pharmacovigilance

GLP-1 drugs have moved far beyond the diabetes and weight-loss conversations where many Americans first encountered them. Patients, providers, researchers, and online communities are now discussing whether Ozempic, Wegovy, Zepbound, and related drugs have applications across a wider set of conditions, from addiction and inflammation to neurological disorders and chronic disease.

Some of those benefits may prove meaningful over time. But the speed and scale of GLP-1 adoption point to a broader challenge for federal health leaders: how do we evaluate safety, efficacy, and real-world outcomes when therapies diffuse quickly across populations, conditions, and care environments.

As therapies are developed, prescribed, and adopted quickly, federal health organizations need better ways to understand how products perform outside the controlled settings in which they were first evaluated.

The Limits of Traditional Pharmacovigilance

Post-market surveillance still relies heavily on formal reporting channels, claims data, electronic health records, clinical notes, and registries. These sources remain essential, but they can miss early signals that emerge in less structured parts of the healthcare ecosystem.

A patient may seek urgent care before filing an adverse event report. A provider may notice a change in medication use before a pattern appears in claims data. A therapy accessed through telehealth or specialty care channels may generate patient experiences that surface online before they are visible in formal systems. In each case, the signal may exist before the surveillance system is able to see it clearly.

That does not mean real-world use should be treated as uncontrolled experimentation. It means pharmacovigilance has to evolve so health leaders can learn from real-world evidence in a more structured and timely way, protect patients, and support the safe adoption of promising therapies.

Better Signal Detection Requires Broader Data

Modern pharmacovigilance depends on signal detection: the ability to identify patterns of emerging adverse events, surrogate markers of safety concerns, unexpected benefits, or areas requiring closer study.

A more modern approach may require broader sources of real-world data and evidence. These could include care-seeking behavior, absenteeism from school or work, changes in normal daily activities, supply chain trends, self-treatment patterns, prescribing patterns, pharmacy activity, telehealth activity, and other indicators that can serve as early proxies for changes in patient experience or potential safety concerns.

The challenge is not collecting more data for its own sake. It is determining which data elements are relevant, how they should be validated, and how to distinguish meaningful signals from noise. AI and advanced analytics can help identify patterns across large, complex, and fragmented data sets. But strong pharmacovigilance still requires clinical domain expertise, scientific rigor, privacy protection, bias monitoring, and clear governance.

Used responsibly, broader data and better analytics can help federal health organizations move from a primarily reactive model toward one that supports earlier hypothesis generation, faster signal triage, and more informed public health action.

Real-Time Data Is Becoming Part of the Regulatory Conversation

Federal health agencies are already exploring faster, more modern ways to use data. The FDA has announced a pilot using cloud and AI to monitor clinical trial data in real time, showing how faster access to signals can support regulatory decision-making on product safety and efficacy.

That pilot is not the same as post-market surveillance, and clinical trial data is governed by strict rules around monitoring and analysis. But it reinforces a broader direction: health agencies are looking for ways to make data available sooner without compromising safety, privacy, or scientific integrity.

As real-world use becomes more complex, pharmacovigilance systems will need to incorporate a wider range of data sources, evaluate signals earlier, and support action across regulators, researchers, clinicians, industry, and technology partners.

What Federal Health Leaders Should Consider Next

Federal health organizations should consider three practical steps.

• First, strengthen real-world signal detection by connecting adverse event reporting with broader sources of real-world data and evidence. This includes identifying which non-traditional indicators may provide useful early insight, where they can be responsibly applied, and how they should be validated.
• Second, apply AI and analytics in ways that help experts identify, prioritize, and evaluate potential signals, rather than replacing scientific review. AI can accelerate pattern detection, but clinical and regulatory judgment must remain central to determining what a signal means and what action should follow.
• Third, create clearer forums for collaboration among regulators, researchers, clinicians, technology partners, and industry. A more proactive model will require shared standards for data quality, privacy, transparency, auditability, interoperability, and governance.

The GLP-1 moment is not only a story about one class of drugs. It is a reminder that pharmacovigilance must keep pace with the speed at which novel therapies move through clinical testing and into real-world use. The future of drug safety will depend on the ability to see signals earlier, interpret them responsibly, and turn fragmented real-world information into better public health decisions.