Antidepressant use in the U.S. has been climbing for years, and so has the length of time patients stay on these medications. That trend recently became a flashpoint: In May 2026, HHS Secretary Robert F. Kennedy Jr. announced a plan aimed at curbing what he describes as a “dependency crisis driven by overmedicalization,” with new training, billing guidance, and clinical resources designed to help providers safely taper patients off psychiatric medications.
The plan was met with pushback from psychiatrists, who called it an oversimplification of a far more complex problem. The American Psychiatric Association’s president noted that too many patients still cannot access timely, comprehensive mental health care in the first place. Mental health advocates and clinicians alike were quick to point out that antidepressants remain effective, evidence-based treatments for millions of people, even as they agreed that providers and patients need better tools and information to navigate tapering and discontinuation when it is clinically appropriate.
What both sides of this debate seem to agree on is this: There isn’t enough real-world data on what actually happens to patients during and after antidepressant treatment. How long are people really staying on these medications? What happens when they stop? Those are the questions we set out to answer in a new study to be presented at the International Society for Pharmacoepidemiology (ISPE) Annual Conference, using NorstellaLinQ real-world data and Panalgo’s Instant Health Data (IHD) platform.
The Data Gap Behind a Political Flashpoint
Real-world evidence on how long people actually stay on antidepressants in the U.S., and what happens, physically and symptomatically, when they discontinue, is limited.
That gap matters because the data needed to answer it doesn’t sit neatly in any single source. Regular antidepressant use is best tracked through closed claims, but the withdrawal symptoms that matter most to patients, like fatigue, insomnia, and flu-like symptoms, are often not severe enough to generate a claim or a hospitalization. Patients experience them and describe them to a provider, but that conversation typically only shows up in the clinical narrative, not in structured EHR fields. Capturing it requires data curation applied to unstructured clinical notes, which is close to the only way to observe this kind of patient experience outside of a prospective questionnaire-based study.
Tracking Patients From First Prescription to Discontinuation
Using NorstellaLinQ data and IHD, we identified patients with regular, habitual antidepressant use, rather than short-term or one-off prescriptions, and tracked them through continuous treatment. We followed each patient to their first instance of discontinuation, and from there, used data curation to mine clinical notes for documented withdrawal symptoms following that discontinuation date.
The results showed that treatment duration is associated with distinct withdrawal symptom profiles: Longer-term users reported a higher prevalence of fatigue, while shorter-term users were more likely to report flu-like symptoms. Future research, we believe, should dig further into how these patterns vary by drug class and during the titration period itself, not just after full discontinuation.
This kind of evidence gives providers something concrete to share with patients before they begin tapering: realistic expectations for what their bodies may experience, and what’s a normal versus an abnormal response. It’s exactly the kind of information that can support the informed consent and shared decision-making that HHS’s own plan calls for, without taking a position on whether any individual patient should or shouldn’t be on antidepressants. The goal isn’t to tell patients to get off their medication. It’s to make sure that when patients and providers decide together to do so, they’re doing it with real information instead of guesswork.
The goal isn’t to tell patients to get off their medication. It’s to make sure that when patients and providers decide together to do so, they’re doing it with real information instead of guesswork.
Why Connected Real-World Data Made the Difference
Neither claims data nor structured EHR data alone could have answered this question. Regular antidepressant use is best measured through closed claims, but withdrawal symptoms live almost exclusively in clinical notes. Capturing both required a data source large enough and rich enough to bridge that gap, which is what made NorstellaLinQ’s combination of structured EHR data and unstructured clinical notes essential to the study design.
Using IHD also meant the analysis itself moved fast. Designing the study took time, but once the cohort and methodology were defined, running the analysis took about a week. That speed allowed for rapid iteration, testing different definitions of “regular use” and discontinuation against the data, rather than getting locked into a single approach before seeing how the data actually behaved. It’s also the foundation for a natural next step: a linked study pairing closed claims with symptom data for a more precise measure of exposure alongside the symptom data already captured here.
Building the Evidence Base Patients and Providers Need
When it comes to RFK Jr.’s plan—which likely isn’t going away—there’s truth to many of the claims being made: antidepressant use has increased, conversations between patients and providers about discontinuation remain rare, and psychiatric medications are still effective, necessary treatments for millions of people. What’s been missing is the real-world evidence to inform that conversation rather than oversimplify it.
This study is a first step toward that evidence base, not a final word on it. As this research continues, the goal stays the same: Give patients and providers the information they need to make safe, informed decisions together.
Contact us to find out how NorstellaLinQ and IHD can help you generate real-world evidence to make decisions, faster.
